/[pcre]/code/trunk/ChangeLog
ViewVC logotype

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1709 - (show annotations)
Tue Aug 1 08:29:13 2017 UTC (2 years, 1 month ago) by zherczeg
File size: 293726 byte(s)
Minor JIT compiler update.
1 ChangeLog for PCRE
2 ------------------
3
4 Note that the PCRE 8.xx series (PCRE1) is now in a bugfix-only state. All
5 development is happening in the PCRE2 10.xx series.
6
7 Version 8.42 xx-xxx-2017
8 ------------------------
9
10 1. Fixed a MIPS issue in the JIT compiler reported by Joshua Kinard.
11
12
13 Version 8.41 05-July-2017
14 -------------------------
15
16 1. Fixed typo in CMakeLists.txt (wrong number of arguments for
17 PCRE_STATIC_RUNTIME (affects MSVC only).
18
19 2. Issue 1 for 8.40 below was not correctly fixed. If pcregrep in multiline
20 mode with --only-matching matched several lines, it restarted scanning at the
21 next line instead of moving on to the end of the matched string, which can be
22 several lines after the start.
23
24 3. Fix a missing else in the JIT compiler reported by 'idaifish'.
25
26 4. A (?# style comment is now ignored between a basic quantifier and a
27 following '+' or '?' (example: /X+(?#comment)?Y/.
28
29 5. Avoid use of a potentially overflowing buffer in pcregrep (patch by Petr
30 Pisar).
31
32 6. Fuzzers have reported issues in pcretest. These are NOT serious (it is,
33 after all, just a test program). However, to stop the reports, some easy ones
34 are fixed:
35
36 (a) Check for values < 256 when calling isprint() in pcretest.
37 (b) Give an error for too big a number after \O.
38
39 7. In the 32-bit library in non-UTF mode, an attempt to find a Unicode
40 property for a character with a code point greater than 0x10ffff (the Unicode
41 maximum) caused a crash.
42
43 8. The alternative matching function, pcre_dfa_exec() misbehaved if it
44 encountered a character class with a possessive repeat, for example [a-f]{3}+.
45
46 9. When pcretest called pcre_copy_substring() in 32-bit mode, it set the buffer
47 length incorrectly, which could result in buffer overflow.
48
49 10. Remove redundant line of code (accidentally left in ages ago).
50
51 11. Applied C++ patch from Irfan Adilovic to guard 'using std::' directives
52 with namespace pcrecpp (Bugzilla #2084).
53
54 12. Remove a duplication typo in pcre_tables.c.
55
56 13. Fix returned offsets from regexec() when REG_STARTEND is used with a
57 starting offset greater than zero.
58
59
60 Version 8.40 11-January-2017
61 ----------------------------
62
63 1. Using -o with -M in pcregrep could cause unnecessary repeated output when
64 the match extended over a line boundary.
65
66 2. Applied Chris Wilson's second patch (Bugzilla #1681) to CMakeLists.txt for
67 MSVC static compilation, putting the first patch under a new option.
68
69 3. Fix register overwite in JIT when SSE2 acceleration is enabled.
70
71 4. Ignore "show all captures" (/=) for DFA matching.
72
73 5. Fix JIT unaligned accesses on x86. Patch by Marc Mutz.
74
75 6. In any wide-character mode (8-bit UTF or any 16-bit or 32-bit mode),
76 without PCRE_UCP set, a negative character type such as \D in a positive
77 class should cause all characters greater than 255 to match, whatever else
78 is in the class. There was a bug that caused this not to happen if a
79 Unicode property item was added to such a class, for example [\D\P{Nd}] or
80 [\W\pL].
81
82 7. When pcretest was outputing information from a callout, the caret indicator
83 for the current position in the subject line was incorrect if it was after
84 an escape sequence for a character whose code point was greater than
85 \x{ff}.
86
87 8. A pattern such as (?<RA>abc)(?(R)xyz) was incorrectly compiled such that
88 the conditional was interpreted as a reference to capturing group 1 instead
89 of a test for recursion. Any group whose name began with R was
90 misinterpreted in this way. (The reference interpretation should only
91 happen if the group's name is precisely "R".)
92
93 9. A number of bugs have been mended relating to match start-up optimizations
94 when the first thing in a pattern is a positive lookahead. These all
95 applied only when PCRE_NO_START_OPTIMIZE was *not* set:
96
97 (a) A pattern such as (?=.*X)X$ was incorrectly optimized as if it needed
98 both an initial 'X' and a following 'X'.
99 (b) Some patterns starting with an assertion that started with .* were
100 incorrectly optimized as having to match at the start of the subject or
101 after a newline. There are cases where this is not true, for example,
102 (?=.*[A-Z])(?=.{8,16})(?!.*[\s]) matches after the start in lines that
103 start with spaces. Starting .* in an assertion is no longer taken as an
104 indication of matching at the start (or after a newline).
105
106
107 Version 8.39 14-June-2016
108 -------------------------
109
110 1. If PCRE_AUTO_CALLOUT was set on a pattern that had a (?# comment between
111 an item and its qualifier (for example, A(?#comment)?B) pcre_compile()
112 misbehaved. This bug was found by the LLVM fuzzer.
113
114 2. Similar to the above, if an isolated \E was present between an item and its
115 qualifier when PCRE_AUTO_CALLOUT was set, pcre_compile() misbehaved. This
116 bug was found by the LLVM fuzzer.
117
118 3. Further to 8.38/46, negated classes such as [^[:^ascii:]\d] were also not
119 working correctly in UCP mode.
120
121 4. The POSIX wrapper function regexec() crashed if the option REG_STARTEND
122 was set when the pmatch argument was NULL. It now returns REG_INVARG.
123
124 5. Allow for up to 32-bit numbers in the ordin() function in pcregrep.
125
126 6. An empty \Q\E sequence between an item and its qualifier caused
127 pcre_compile() to misbehave when auto callouts were enabled. This bug was
128 found by the LLVM fuzzer.
129
130 7. If a pattern that was compiled with PCRE_EXTENDED started with white
131 space or a #-type comment that was followed by (?-x), which turns off
132 PCRE_EXTENDED, and there was no subsequent (?x) to turn it on again,
133 pcre_compile() assumed that (?-x) applied to the whole pattern and
134 consequently mis-compiled it. This bug was found by the LLVM fuzzer.
135
136 8. A call of pcre_copy_named_substring() for a named substring whose number
137 was greater than the space in the ovector could cause a crash.
138
139 9. Yet another buffer overflow bug involved duplicate named groups with a
140 group that reset capture numbers (compare 8.38/7 below). Once again, I have
141 just allowed for more memory, even if not needed. (A proper fix is
142 implemented in PCRE2, but it involves a lot of refactoring.)
143
144 10. pcre_get_substring_list() crashed if the use of \K in a match caused the
145 start of the match to be earlier than the end.
146
147 11. Migrating appropriate PCRE2 JIT improvements to PCRE.
148
149 12. A pattern such as /(?<=((?C)0))/, which has a callout inside a lookbehind
150 assertion, caused pcretest to generate incorrect output, and also to read
151 uninitialized memory (detected by ASAN or valgrind).
152
153 13. A pattern that included (*ACCEPT) in the middle of a sufficiently deeply
154 nested set of parentheses of sufficient size caused an overflow of the
155 compiling workspace (which was diagnosed, but of course is not desirable).
156
157 14. And yet another buffer overflow bug involving duplicate named groups, this
158 time nested, with a nested back reference. Yet again, I have just allowed
159 for more memory, because anything more needs all the refactoring that has
160 been done for PCRE2. An example pattern that provoked this bug is:
161 /((?J)(?'R'(?'R'(?'R'(?'R'(?'R'(?|(\k'R'))))))))/ and the bug was
162 registered as CVE-2016-1283.
163
164 15. pcretest went into a loop if global matching was requested with an ovector
165 size less than 2. It now gives an error message. This bug was found by
166 afl-fuzz.
167
168 16. An invalid pattern fragment such as (?(?C)0 was not diagnosing an error
169 ("assertion expected") when (?(?C) was not followed by an opening
170 parenthesis.
171
172 17. Fixed typo ("&&" for "&") in pcre_study(). Fortunately, this could not
173 actually affect anything, by sheer luck.
174
175 18. Applied Chris Wilson's patch (Bugzilla #1681) to CMakeLists.txt for MSVC
176 static compilation.
177
178 19. Modified the RunTest script to incorporate a valgrind suppressions file so
179 that certain errors, provoked by the SSE2 instruction set when JIT is used,
180 are ignored.
181
182 20. A racing condition is fixed in JIT reported by Mozilla.
183
184 21. Minor code refactor to avoid "array subscript is below array bounds"
185 compiler warning.
186
187 22. Minor code refactor to avoid "left shift of negative number" warning.
188
189 23. Fix typo causing compile error when 16- or 32-bit JIT is compiled without
190 UCP support.
191
192 24. Refactor to avoid compiler warnings in pcrecpp.cc.
193
194 25. Refactor to fix a typo in pcre_jit_test.c
195
196 26. Patch to support compiling pcrecpp.cc with Intel compiler.
197
198
199 Version 8.38 23-November-2015
200 -----------------------------
201
202 1. If a group that contained a recursive back reference also contained a
203 forward reference subroutine call followed by a non-forward-reference
204 subroutine call, for example /.((?2)(?R)\1)()/, pcre_compile() failed to
205 compile correct code, leading to undefined behaviour or an internally
206 detected error. This bug was discovered by the LLVM fuzzer.
207
208 2. Quantification of certain items (e.g. atomic back references) could cause
209 incorrect code to be compiled when recursive forward references were
210 involved. For example, in this pattern: /(?1)()((((((\1++))\x85)+)|))/.
211 This bug was discovered by the LLVM fuzzer.
212
213 3. A repeated conditional group whose condition was a reference by name caused
214 a buffer overflow if there was more than one group with the given name.
215 This bug was discovered by the LLVM fuzzer.
216
217 4. A recursive back reference by name within a group that had the same name as
218 another group caused a buffer overflow. For example:
219 /(?J)(?'d'(?'d'\g{d}))/. This bug was discovered by the LLVM fuzzer.
220
221 5. A forward reference by name to a group whose number is the same as the
222 current group, for example in this pattern: /(?|(\k'Pm')|(?'Pm'))/, caused
223 a buffer overflow at compile time. This bug was discovered by the LLVM
224 fuzzer.
225
226 6. A lookbehind assertion within a set of mutually recursive subpatterns could
227 provoke a buffer overflow. This bug was discovered by the LLVM fuzzer.
228
229 7. Another buffer overflow bug involved duplicate named groups with a
230 reference between their definition, with a group that reset capture
231 numbers, for example: /(?J:(?|(?'R')(\k'R')|((?'R'))))/. This has been
232 fixed by always allowing for more memory, even if not needed. (A proper fix
233 is implemented in PCRE2, but it involves more refactoring.)
234
235 8. There was no check for integer overflow in subroutine calls such as (?123).
236
237 9. The table entry for \l in EBCDIC environments was incorrect, leading to its
238 being treated as a literal 'l' instead of causing an error.
239
240 10. There was a buffer overflow if pcre_exec() was called with an ovector of
241 size 1. This bug was found by american fuzzy lop.
242
243 11. If a non-capturing group containing a conditional group that could match
244 an empty string was repeated, it was not identified as matching an empty
245 string itself. For example: /^(?:(?(1)x|)+)+$()/.
246
247 12. In an EBCDIC environment, pcretest was mishandling the escape sequences
248 \a and \e in test subject lines.
249
250 13. In an EBCDIC environment, \a in a pattern was converted to the ASCII
251 instead of the EBCDIC value.
252
253 14. The handling of \c in an EBCDIC environment has been revised so that it is
254 now compatible with the specification in Perl's perlebcdic page.
255
256 15. The EBCDIC character 0x41 is a non-breaking space, equivalent to 0xa0 in
257 ASCII/Unicode. This has now been added to the list of characters that are
258 recognized as white space in EBCDIC.
259
260 16. When PCRE was compiled without UCP support, the use of \p and \P gave an
261 error (correctly) when used outside a class, but did not give an error
262 within a class.
263
264 17. \h within a class was incorrectly compiled in EBCDIC environments.
265
266 18. A pattern with an unmatched closing parenthesis that contained a backward
267 assertion which itself contained a forward reference caused buffer
268 overflow. And example pattern is: /(?=di(?<=(?1))|(?=(.))))/.
269
270 19. JIT should return with error when the compiled pattern requires more stack
271 space than the maximum.
272
273 20. A possessively repeated conditional group that could match an empty string,
274 for example, /(?(R))*+/, was incorrectly compiled.
275
276 21. Fix infinite recursion in the JIT compiler when certain patterns such as
277 /(?:|a|){100}x/ are analysed.
278
279 22. Some patterns with character classes involving [: and \\ were incorrectly
280 compiled and could cause reading from uninitialized memory or an incorrect
281 error diagnosis.
282
283 23. Pathological patterns containing many nested occurrences of [: caused
284 pcre_compile() to run for a very long time.
285
286 24. A conditional group with only one branch has an implicit empty alternative
287 branch and must therefore be treated as potentially matching an empty
288 string.
289
290 25. If (?R was followed by - or + incorrect behaviour happened instead of a
291 diagnostic.
292
293 26. Arrange to give up on finding the minimum matching length for overly
294 complex patterns.
295
296 27. Similar to (4) above: in a pattern with duplicated named groups and an
297 occurrence of (?| it is possible for an apparently non-recursive back
298 reference to become recursive if a later named group with the relevant
299 number is encountered. This could lead to a buffer overflow. Wen Guanxing
300 from Venustech ADLAB discovered this bug.
301
302 28. If pcregrep was given the -q option with -c or -l, or when handling a
303 binary file, it incorrectly wrote output to stdout.
304
305 29. The JIT compiler did not restore the control verb head in case of *THEN
306 control verbs. This issue was found by Karl Skomski with a custom LLVM
307 fuzzer.
308
309 30. Error messages for syntax errors following \g and \k were giving inaccurate
310 offsets in the pattern.
311
312 31. Added a check for integer overflow in conditions (?(<digits>) and
313 (?(R<digits>). This omission was discovered by Karl Skomski with the LLVM
314 fuzzer.
315
316 32. Handling recursive references such as (?2) when the reference is to a group
317 later in the pattern uses code that is very hacked about and error-prone.
318 It has been re-written for PCRE2. Here in PCRE1, a check has been added to
319 give an internal error if it is obvious that compiling has gone wrong.
320
321 33. The JIT compiler should not check repeats after a {0,1} repeat byte code.
322 This issue was found by Karl Skomski with a custom LLVM fuzzer.
323
324 34. The JIT compiler should restore the control chain for empty possessive
325 repeats. This issue was found by Karl Skomski with a custom LLVM fuzzer.
326
327 35. Match limit check added to JIT recursion. This issue was found by Karl
328 Skomski with a custom LLVM fuzzer.
329
330 36. Yet another case similar to 27 above has been circumvented by an
331 unconditional allocation of extra memory. This issue is fixed "properly" in
332 PCRE2 by refactoring the way references are handled. Wen Guanxing
333 from Venustech ADLAB discovered this bug.
334
335 37. Fix two assertion fails in JIT. These issues were found by Karl Skomski
336 with a custom LLVM fuzzer.
337
338 38. Fixed a corner case of range optimization in JIT.
339
340 39. An incorrect error "overran compiling workspace" was given if there were
341 exactly enough group forward references such that the last one extended
342 into the workspace safety margin. The next one would have expanded the
343 workspace. The test for overflow was not including the safety margin.
344
345 40. A match limit issue is fixed in JIT which was found by Karl Skomski
346 with a custom LLVM fuzzer.
347
348 41. Remove the use of /dev/null in testdata/testinput2, because it doesn't
349 work under Windows. (Why has it taken so long for anyone to notice?)
350
351 42. In a character class such as [\W\p{Any}] where both a negative-type escape
352 ("not a word character") and a property escape were present, the property
353 escape was being ignored.
354
355 43. Fix crash caused by very long (*MARK) or (*THEN) names.
356
357 44. A sequence such as [[:punct:]b] that is, a POSIX character class followed
358 by a single ASCII character in a class item, was incorrectly compiled in
359 UCP mode. The POSIX class got lost, but only if the single character
360 followed it.
361
362 45. [:punct:] in UCP mode was matching some characters in the range 128-255
363 that should not have been matched.
364
365 46. If [:^ascii:] or [:^xdigit:] or [:^cntrl:] are present in a non-negated
366 class, all characters with code points greater than 255 are in the class.
367 When a Unicode property was also in the class (if PCRE_UCP is set, escapes
368 such as \w are turned into Unicode properties), wide characters were not
369 correctly handled, and could fail to match.
370
371
372 Version 8.37 28-April-2015
373 --------------------------
374
375 1. When an (*ACCEPT) is triggered inside capturing parentheses, it arranges
376 for those parentheses to be closed with whatever has been captured so far.
377 However, it was failing to mark any other groups between the hightest
378 capture so far and the currrent group as "unset". Thus, the ovector for
379 those groups contained whatever was previously there. An example is the
380 pattern /(x)|((*ACCEPT))/ when matched against "abcd".
381
382 2. If an assertion condition was quantified with a minimum of zero (an odd
383 thing to do, but it happened), SIGSEGV or other misbehaviour could occur.
384
385 3. If a pattern in pcretest input had the P (POSIX) modifier followed by an
386 unrecognized modifier, a crash could occur.
387
388 4. An attempt to do global matching in pcretest with a zero-length ovector
389 caused a crash.
390
391 5. Fixed a memory leak during matching that could occur for a subpattern
392 subroutine call (recursive or otherwise) if the number of captured groups
393 that had to be saved was greater than ten.
394
395 6. Catch a bad opcode during auto-possessification after compiling a bad UTF
396 string with NO_UTF_CHECK. This is a tidyup, not a bug fix, as passing bad
397 UTF with NO_UTF_CHECK is documented as having an undefined outcome.
398
399 7. A UTF pattern containing a "not" match of a non-ASCII character and a
400 subroutine reference could loop at compile time. Example: /[^\xff]((?1))/.
401
402 8. When a pattern is compiled, it remembers the highest back reference so that
403 when matching, if the ovector is too small, extra memory can be obtained to
404 use instead. A conditional subpattern whose condition is a check on a
405 capture having happened, such as, for example in the pattern
406 /^(?:(a)|b)(?(1)A|B)/, is another kind of back reference, but it was not
407 setting the highest backreference number. This mattered only if pcre_exec()
408 was called with an ovector that was too small to hold the capture, and there
409 was no other kind of back reference (a situation which is probably quite
410 rare). The effect of the bug was that the condition was always treated as
411 FALSE when the capture could not be consulted, leading to a incorrect
412 behaviour by pcre_exec(). This bug has been fixed.
413
414 9. A reference to a duplicated named group (either a back reference or a test
415 for being set in a conditional) that occurred in a part of the pattern where
416 PCRE_DUPNAMES was not set caused the amount of memory needed for the pattern
417 to be incorrectly calculated, leading to overwriting.
418
419 10. A mutually recursive set of back references such as (\2)(\1) caused a
420 segfault at study time (while trying to find the minimum matching length).
421 The infinite loop is now broken (with the minimum length unset, that is,
422 zero).
423
424 11. If an assertion that was used as a condition was quantified with a minimum
425 of zero, matching went wrong. In particular, if the whole group had
426 unlimited repetition and could match an empty string, a segfault was
427 likely. The pattern (?(?=0)?)+ is an example that caused this. Perl allows
428 assertions to be quantified, but not if they are being used as conditions,
429 so the above pattern is faulted by Perl. PCRE has now been changed so that
430 it also rejects such patterns.
431
432 12. A possessive capturing group such as (a)*+ with a minimum repeat of zero
433 failed to allow the zero-repeat case if pcre2_exec() was called with an
434 ovector too small to capture the group.
435
436 13. Fixed two bugs in pcretest that were discovered by fuzzing and reported by
437 Red Hat Product Security:
438
439 (a) A crash if /K and /F were both set with the option to save the compiled
440 pattern.
441
442 (b) Another crash if the option to print captured substrings in a callout
443 was combined with setting a null ovector, for example \O\C+ as a subject
444 string.
445
446 14. A pattern such as "((?2){0,1999}())?", which has a group containing a
447 forward reference repeated a large (but limited) number of times within a
448 repeated outer group that has a zero minimum quantifier, caused incorrect
449 code to be compiled, leading to the error "internal error:
450 previously-checked referenced subpattern not found" when an incorrect
451 memory address was read. This bug was reported as "heap overflow",
452 discovered by Kai Lu of Fortinet's FortiGuard Labs and given the CVE number
453 CVE-2015-2325.
454
455 23. A pattern such as "((?+1)(\1))/" containing a forward reference subroutine
456 call within a group that also contained a recursive back reference caused
457 incorrect code to be compiled. This bug was reported as "heap overflow",
458 discovered by Kai Lu of Fortinet's FortiGuard Labs, and given the CVE
459 number CVE-2015-2326.
460
461 24. Computing the size of the JIT read-only data in advance has been a source
462 of various issues, and new ones are still appear unfortunately. To fix
463 existing and future issues, size computation is eliminated from the code,
464 and replaced by on-demand memory allocation.
465
466 25. A pattern such as /(?i)[A-`]/, where characters in the other case are
467 adjacent to the end of the range, and the range contained characters with
468 more than one other case, caused incorrect behaviour when compiled in UTF
469 mode. In that example, the range a-j was left out of the class.
470
471 26. Fix JIT compilation of conditional blocks, which assertion
472 is converted to (*FAIL). E.g: /(?(?!))/.
473
474 27. The pattern /(?(?!)^)/ caused references to random memory. This bug was
475 discovered by the LLVM fuzzer.
476
477 28. The assertion (?!) is optimized to (*FAIL). This was not handled correctly
478 when this assertion was used as a condition, for example (?(?!)a|b). In
479 pcre2_match() it worked by luck; in pcre2_dfa_match() it gave an incorrect
480 error about an unsupported item.
481
482 29. For some types of pattern, for example /Z*(|d*){216}/, the auto-
483 possessification code could take exponential time to complete. A recursion
484 depth limit of 1000 has been imposed to limit the resources used by this
485 optimization.
486
487 30. A pattern such as /(*UTF)[\S\V\H]/, which contains a negated special class
488 such as \S in non-UCP mode, explicit wide characters (> 255) can be ignored
489 because \S ensures they are all in the class. The code for doing this was
490 interacting badly with the code for computing the amount of space needed to
491 compile the pattern, leading to a buffer overflow. This bug was discovered
492 by the LLVM fuzzer.
493
494 31. A pattern such as /((?2)+)((?1))/ which has mutual recursion nested inside
495 other kinds of group caused stack overflow at compile time. This bug was
496 discovered by the LLVM fuzzer.
497
498 32. A pattern such as /(?1)(?#?'){8}(a)/ which had a parenthesized comment
499 between a subroutine call and its quantifier was incorrectly compiled,
500 leading to buffer overflow or other errors. This bug was discovered by the
501 LLVM fuzzer.
502
503 33. The illegal pattern /(?(?<E>.*!.*)?)/ was not being diagnosed as missing an
504 assertion after (?(. The code was failing to check the character after
505 (?(?< for the ! or = that would indicate a lookbehind assertion. This bug
506 was discovered by the LLVM fuzzer.
507
508 34. A pattern such as /X((?2)()*+){2}+/ which has a possessive quantifier with
509 a fixed maximum following a group that contains a subroutine reference was
510 incorrectly compiled and could trigger buffer overflow. This bug was
511 discovered by the LLVM fuzzer.
512
513 35. A mutual recursion within a lookbehind assertion such as (?<=((?2))((?1)))
514 caused a stack overflow instead of the diagnosis of a non-fixed length
515 lookbehind assertion. This bug was discovered by the LLVM fuzzer.
516
517 36. The use of \K in a positive lookbehind assertion in a non-anchored pattern
518 (e.g. /(?<=\Ka)/) could make pcregrep loop.
519
520 37. There was a similar problem to 36 in pcretest for global matches.
521
522 38. If a greedy quantified \X was preceded by \C in UTF mode (e.g. \C\X*),
523 and a subsequent item in the pattern caused a non-match, backtracking over
524 the repeated \X did not stop, but carried on past the start of the subject,
525 causing reference to random memory and/or a segfault. There were also some
526 other cases where backtracking after \C could crash. This set of bugs was
527 discovered by the LLVM fuzzer.
528
529 39. The function for finding the minimum length of a matching string could take
530 a very long time if mutual recursion was present many times in a pattern,
531 for example, /((?2){73}(?2))((?1))/. A better mutual recursion detection
532 method has been implemented. This infelicity was discovered by the LLVM
533 fuzzer.
534
535 40. Static linking against the PCRE library using the pkg-config module was
536 failing on missing pthread symbols.
537
538
539 Version 8.36 26-September-2014
540 ------------------------------
541
542 1. Got rid of some compiler warnings in the C++ modules that were shown up by
543 -Wmissing-field-initializers and -Wunused-parameter.
544
545 2. The tests for quantifiers being too big (greater than 65535) were being
546 applied after reading the number, and stupidly assuming that integer
547 overflow would give a negative number. The tests are now applied as the
548 numbers are read.
549
550 3. Tidy code in pcre_exec.c where two branches that used to be different are
551 now the same.
552
553 4. The JIT compiler did not generate match limit checks for certain
554 bracketed expressions with quantifiers. This may lead to exponential
555 backtracking, instead of returning with PCRE_ERROR_MATCHLIMIT. This
556 issue should be resolved now.
557
558 5. Fixed an issue, which occures when nested alternatives are optimized
559 with table jumps.
560
561 6. Inserted two casts and changed some ints to size_t in the light of some
562 reported 64-bit compiler warnings (Bugzilla 1477).
563
564 7. Fixed a bug concerned with zero-minimum possessive groups that could match
565 an empty string, which sometimes were behaving incorrectly in the
566 interpreter (though correctly in the JIT matcher). This pcretest input is
567 an example:
568
569 '\A(?:[^"]++|"(?:[^"]*+|"")*+")++'
570 NON QUOTED "QUOT""ED" AFTER "NOT MATCHED
571
572 the interpreter was reporting a match of 'NON QUOTED ' only, whereas the
573 JIT matcher and Perl both matched 'NON QUOTED "QUOT""ED" AFTER '. The test
574 for an empty string was breaking the inner loop and carrying on at a lower
575 level, when possessive repeated groups should always return to a higher
576 level as they have no backtrack points in them. The empty string test now
577 occurs at the outer level.
578
579 8. Fixed a bug that was incorrectly auto-possessifying \w+ in the pattern
580 ^\w+(?>\s*)(?<=\w) which caused it not to match "test test".
581
582 9. Give a compile-time error for \o{} (as Perl does) and for \x{} (which Perl
583 doesn't).
584
585 10. Change 8.34/15 introduced a bug that caused the amount of memory needed
586 to hold a pattern to be incorrectly computed (too small) when there were
587 named back references to duplicated names. This could cause "internal
588 error: code overflow" or "double free or corruption" or other memory
589 handling errors.
590
591 11. When named subpatterns had the same prefixes, back references could be
592 confused. For example, in this pattern:
593
594 /(?P<Name>a)?(?P<Name2>b)?(?(<Name>)c|d)*l/
595
596 the reference to 'Name' was incorrectly treated as a reference to a
597 duplicate name.
598
599 12. A pattern such as /^s?c/mi8 where the optional character has more than
600 one "other case" was incorrectly compiled such that it would only try to
601 match starting at "c".
602
603 13. When a pattern starting with \s was studied, VT was not included in the
604 list of possible starting characters; this should have been part of the
605 8.34/18 patch.
606
607 14. If a character class started [\Qx]... where x is any character, the class
608 was incorrectly terminated at the ].
609
610 15. If a pattern that started with a caseless match for a character with more
611 than one "other case" was studied, PCRE did not set up the starting code
612 unit bit map for the list of possible characters. Now it does. This is an
613 optimization improvement, not a bug fix.
614
615 16. The Unicode data tables have been updated to Unicode 7.0.0.
616
617 17. Fixed a number of memory leaks in pcregrep.
618
619 18. Avoid a compiler warning (from some compilers) for a function call with
620 a cast that removes "const" from an lvalue by using an intermediate
621 variable (to which the compiler does not object).
622
623 19. Incorrect code was compiled if a group that contained an internal recursive
624 back reference was optional (had quantifier with a minimum of zero). This
625 example compiled incorrect code: /(((a\2)|(a*)\g<-1>))*/ and other examples
626 caused segmentation faults because of stack overflows at compile time.
627
628 20. A pattern such as /((?(R)a|(?1)))+/, which contains a recursion within a
629 group that is quantified with an indefinite repeat, caused a compile-time
630 loop which used up all the system stack and provoked a segmentation fault.
631 This was not the same bug as 19 above.
632
633 21. Add PCRECPP_EXP_DECL declaration to operator<< in pcre_stringpiece.h.
634 Patch by Mike Frysinger.
635
636
637 Version 8.35 04-April-2014
638 --------------------------
639
640 1. A new flag is set, when property checks are present in an XCLASS.
641 When this flag is not set, PCRE can perform certain optimizations
642 such as studying these XCLASS-es.
643
644 2. The auto-possessification of character sets were improved: a normal
645 and an extended character set can be compared now. Furthermore
646 the JIT compiler optimizes more character set checks.
647
648 3. Got rid of some compiler warnings for potentially uninitialized variables
649 that show up only when compiled with -O2.
650
651 4. A pattern such as (?=ab\K) that uses \K in an assertion can set the start
652 of a match later then the end of the match. The pcretest program was not
653 handling the case sensibly - it was outputting from the start to the next
654 binary zero. It now reports this situation in a message, and outputs the
655 text from the end to the start.
656
657 5. Fast forward search is improved in JIT. Instead of the first three
658 characters, any three characters with fixed position can be searched.
659 Search order: first, last, middle.
660
661 6. Improve character range checks in JIT. Characters are read by an inprecise
662 function now, which returns with an unknown value if the character code is
663 above a certain threshold (e.g: 256). The only limitation is that the value
664 must be bigger than the threshold as well. This function is useful when
665 the characters above the threshold are handled in the same way.
666
667 7. The macros whose names start with RAWUCHAR are placeholders for a future
668 mode in which only the bottom 21 bits of 32-bit data items are used. To
669 make this more memorable for those maintaining the code, the names have
670 been changed to start with UCHAR21, and an extensive comment has been added
671 to their definition.
672
673 8. Add missing (new) files sljitNativeTILEGX.c and sljitNativeTILEGX-encoder.c
674 to the export list in Makefile.am (they were accidentally omitted from the
675 8.34 tarball).
676
677 9. The informational output from pcretest used the phrase "starting byte set"
678 which is inappropriate for the 16-bit and 32-bit libraries. As the output
679 for "first char" and "need char" really means "non-UTF-char", I've changed
680 "byte" to "char", and slightly reworded the output. The documentation about
681 these values has also been (I hope) clarified.
682
683 10. Another JIT related optimization: use table jumps for selecting the correct
684 backtracking path, when more than four alternatives are present inside a
685 bracket.
686
687 11. Empty match is not possible, when the minimum length is greater than zero,
688 and there is no \K in the pattern. JIT should avoid empty match checks in
689 such cases.
690
691 12. In a caseless character class with UCP support, when a character with more
692 than one alternative case was not the first character of a range, not all
693 the alternative cases were added to the class. For example, s and \x{17f}
694 are both alternative cases for S: the class [RST] was handled correctly,
695 but [R-T] was not.
696
697 13. The configure.ac file always checked for pthread support when JIT was
698 enabled. This is not used in Windows, so I have put this test inside a
699 check for the presence of windows.h (which was already tested for).
700
701 14. Improve pattern prefix search by a simplified Boyer-Moore algorithm in JIT.
702 The algorithm provides a way to skip certain starting offsets, and usually
703 faster than linear prefix searches.
704
705 15. Change 13 for 8.20 updated RunTest to check for the 'fr' locale as well
706 as for 'fr_FR' and 'french'. For some reason, however, it then used the
707 Windows-specific input and output files, which have 'french' screwed in.
708 So this could never have worked. One of the problems with locales is that
709 they aren't always the same. I have now updated RunTest so that it checks
710 the output of the locale test (test 3) against three different output
711 files, and it allows the test to pass if any one of them matches. With luck
712 this should make the test pass on some versions of Solaris where it was
713 failing. Because of the uncertainty, the script did not used to stop if
714 test 3 failed; it now does. If further versions of a French locale ever
715 come to light, they can now easily be added.
716
717 16. If --with-pcregrep-bufsize was given a non-integer value such as "50K",
718 there was a message during ./configure, but it did not stop. This now
719 provokes an error. The invalid example in README has been corrected.
720 If a value less than the minimum is given, the minimum value has always
721 been used, but now a warning is given.
722
723 17. If --enable-bsr-anycrlf was set, the special 16/32-bit test failed. This
724 was a bug in the test system, which is now fixed. Also, the list of various
725 configurations that are tested for each release did not have one with both
726 16/32 bits and --enable-bar-anycrlf. It now does.
727
728 18. pcretest was missing "-C bsr" for displaying the \R default setting.
729
730 19. Little endian PowerPC systems are supported now by the JIT compiler.
731
732 20. The fast forward newline mechanism could enter to an infinite loop on
733 certain invalid UTF-8 input. Although we don't support these cases
734 this issue can be fixed by a performance optimization.
735
736 21. Change 33 of 8.34 is not sufficient to ensure stack safety because it does
737 not take account if existing stack usage. There is now a new global
738 variable called pcre_stack_guard that can be set to point to an external
739 function to check stack availability. It is called at the start of
740 processing every parenthesized group.
741
742 22. A typo in the code meant that in ungreedy mode the max/min qualifier
743 behaved like a min-possessive qualifier, and, for example, /a{1,3}b/U did
744 not match "ab".
745
746 23. When UTF was disabled, the JIT program reported some incorrect compile
747 errors. These messages are silenced now.
748
749 24. Experimental support for ARM-64 and MIPS-64 has been added to the JIT
750 compiler.
751
752 25. Change all the temporary files used in RunGrepTest to be different to those
753 used by RunTest so that the tests can be run simultaneously, for example by
754 "make -j check".
755
756
757 Version 8.34 15-December-2013
758 -----------------------------
759
760 1. Add pcre[16|32]_jit_free_unused_memory to forcibly free unused JIT
761 executable memory. Patch inspired by Carsten Klein.
762
763 2. ./configure --enable-coverage defined SUPPORT_GCOV in config.h, although
764 this macro is never tested and has no effect, because the work to support
765 coverage involves only compiling and linking options and special targets in
766 the Makefile. The comment in config.h implied that defining the macro would
767 enable coverage support, which is totally false. There was also support for
768 setting this macro in the CMake files (my fault, I just copied it from
769 configure). SUPPORT_GCOV has now been removed.
770
771 3. Make a small performance improvement in strlen16() and strlen32() in
772 pcretest.
773
774 4. Change 36 for 8.33 left some unreachable statements in pcre_exec.c,
775 detected by the Solaris compiler (gcc doesn't seem to be able to diagnose
776 these cases). There was also one in pcretest.c.
777
778 5. Cleaned up a "may be uninitialized" compiler warning in pcre_exec.c.
779
780 6. In UTF mode, the code for checking whether a group could match an empty
781 string (which is used for indefinitely repeated groups to allow for
782 breaking an infinite loop) was broken when the group contained a repeated
783 negated single-character class with a character that occupied more than one
784 data item and had a minimum repetition of zero (for example, [^\x{100}]* in
785 UTF-8 mode). The effect was undefined: the group might or might not be
786 deemed as matching an empty string, or the program might have crashed.
787
788 7. The code for checking whether a group could match an empty string was not
789 recognizing that \h, \H, \v, \V, and \R must match a character.
790
791 8. Implemented PCRE_INFO_MATCH_EMPTY, which yields 1 if the pattern can match
792 an empty string. If it can, pcretest shows this in its information output.
793
794 9. Fixed two related bugs that applied to Unicode extended grapheme clusters
795 that were repeated with a maximizing qualifier (e.g. \X* or \X{2,5}) when
796 matched by pcre_exec() without using JIT:
797
798 (a) If the rest of the pattern did not match after a maximal run of
799 grapheme clusters, the code for backing up to try with fewer of them
800 did not always back up over a full grapheme when characters that do not
801 have the modifier quality were involved, e.g. Hangul syllables.
802
803 (b) If the match point in a subject started with modifier character, and
804 there was no match, the code could incorrectly back up beyond the match
805 point, and potentially beyond the first character in the subject,
806 leading to a segfault or an incorrect match result.
807
808 10. A conditional group with an assertion condition could lead to PCRE
809 recording an incorrect first data item for a match if no other first data
810 item was recorded. For example, the pattern (?(?=ab)ab) recorded "a" as a
811 first data item, and therefore matched "ca" after "c" instead of at the
812 start.
813
814 11. Change 40 for 8.33 (allowing pcregrep to find empty strings) showed up a
815 bug that caused the command "echo a | ./pcregrep -M '|a'" to loop.
816
817 12. The source of pcregrep now includes z/OS-specific code so that it can be
818 compiled for z/OS as part of the special z/OS distribution.
819
820 13. Added the -T and -TM options to pcretest.
821
822 14. The code in pcre_compile.c for creating the table of named capturing groups
823 has been refactored. Instead of creating the table dynamically during the
824 actual compiling pass, the information is remembered during the pre-compile
825 pass (on the stack unless there are more than 20 named groups, in which
826 case malloc() is used) and the whole table is created before the actual
827 compile happens. This has simplified the code (it is now nearly 150 lines
828 shorter) and prepared the way for better handling of references to groups
829 with duplicate names.
830
831 15. A back reference to a named subpattern when there is more than one of the
832 same name now checks them in the order in which they appear in the pattern.
833 The first one that is set is used for the reference. Previously only the
834 first one was inspected. This change makes PCRE more compatible with Perl.
835
836 16. Unicode character properties were updated from Unicode 6.3.0.
837
838 17. The compile-time code for auto-possessification has been refactored, based
839 on a patch by Zoltan Herczeg. It now happens after instead of during
840 compilation. The code is cleaner, and more cases are handled. The option
841 PCRE_NO_AUTO_POSSESS is added for testing purposes, and the -O and /O
842 options in pcretest are provided to set it. It can also be set by
843 (*NO_AUTO_POSSESS) at the start of a pattern.
844
845 18. The character VT has been added to the default ("C" locale) set of
846 characters that match \s and are generally treated as white space,
847 following this same change in Perl 5.18. There is now no difference between
848 "Perl space" and "POSIX space". Whether VT is treated as white space in
849 other locales depends on the locale.
850
851 19. The code for checking named groups as conditions, either for being set or
852 for being recursed, has been refactored (this is related to 14 and 15
853 above). Processing unduplicated named groups should now be as fast at
854 numerical groups, and processing duplicated groups should be faster than
855 before.
856
857 20. Two patches to the CMake build system, by Alexander Barkov:
858
859 (1) Replace the "source" command by "." in CMakeLists.txt because
860 "source" is a bash-ism.
861
862 (2) Add missing HAVE_STDINT_H and HAVE_INTTYPES_H to config-cmake.h.in;
863 without these the CMake build does not work on Solaris.
864
865 21. Perl has changed its handling of \8 and \9. If there is no previously
866 encountered capturing group of those numbers, they are treated as the
867 literal characters 8 and 9 instead of a binary zero followed by the
868 literals. PCRE now does the same.
869
870 22. Following Perl, added \o{} to specify codepoints in octal, making it
871 possible to specify values greater than 0777 and also making them
872 unambiguous.
873
874 23. Perl now gives an error for missing closing braces after \x{... instead of
875 treating the string as literal. PCRE now does the same.
876
877 24. RunTest used to grumble if an inappropriate test was selected explicitly,
878 but just skip it when running all tests. This make it awkward to run ranges
879 of tests when one of them was inappropriate. Now it just skips any
880 inappropriate tests, as it always did when running all tests.
881
882 25. If PCRE_AUTO_CALLOUT and PCRE_UCP were set for a pattern that contained
883 character types such as \d or \w, too many callouts were inserted, and the
884 data that they returned was rubbish.
885
886 26. In UCP mode, \s was not matching two of the characters that Perl matches,
887 namely NEL (U+0085) and MONGOLIAN VOWEL SEPARATOR (U+180E), though they
888 were matched by \h. The code has now been refactored so that the lists of
889 the horizontal and vertical whitespace characters used for \h and \v (which
890 are defined only in one place) are now also used for \s.
891
892 27. Add JIT support for the 64 bit TileGX architecture.
893 Patch by Jiong Wang (Tilera Corporation).
894
895 28. Possessive quantifiers for classes (both explicit and automatically
896 generated) now use special opcodes instead of wrapping in ONCE brackets.
897
898 29. Whereas an item such as A{4}+ ignored the possessivenes of the quantifier
899 (because it's meaningless), this was not happening when PCRE_CASELESS was
900 set. Not wrong, but inefficient.
901
902 30. Updated perltest.pl to add /u (force Unicode mode) when /W (use Unicode
903 properties for \w, \d, etc) is present in a test regex. Otherwise if the
904 test contains no characters greater than 255, Perl doesn't realise it
905 should be using Unicode semantics.
906
907 31. Upgraded the handling of the POSIX classes [:graph:], [:print:], and
908 [:punct:] when PCRE_UCP is set so as to include the same characters as Perl
909 does in Unicode mode.
910
911 32. Added the "forbid" facility to pcretest so that putting tests into the
912 wrong test files can sometimes be quickly detected.
913
914 33. There is now a limit (default 250) on the depth of nesting of parentheses.
915 This limit is imposed to control the amount of system stack used at compile
916 time. It can be changed at build time by --with-parens-nest-limit=xxx or
917 the equivalent in CMake.
918
919 34. Character classes such as [A-\d] or [a-[:digit:]] now cause compile-time
920 errors. Perl warns for these when in warning mode, but PCRE has no facility
921 for giving warnings.
922
923 35. Change 34 for 8.13 allowed quantifiers on assertions, because Perl does.
924 However, this was not working for (?!) because it is optimized to (*FAIL),
925 for which PCRE does not allow quantifiers. The optimization is now disabled
926 when a quantifier follows (?!). I can't see any use for this, but it makes
927 things uniform.
928
929 36. Perl no longer allows group names to start with digits, so I have made this
930 change also in PCRE. It simplifies the code a bit.
931
932 37. In extended mode, Perl ignores spaces before a + that indicates a
933 possessive quantifier. PCRE allowed a space before the quantifier, but not
934 before the possessive +. It now does.
935
936 38. The use of \K (reset reported match start) within a repeated possessive
937 group such as (a\Kb)*+ was not working.
938
939 40. Document that the same character tables must be used at compile time and
940 run time, and that the facility to pass tables to pcre_exec() and
941 pcre_dfa_exec() is for use only with saved/restored patterns.
942
943 41. Applied Jeff Trawick's patch CMakeLists.txt, which "provides two new
944 features for Builds with MSVC:
945
946 1. Support pcre.rc and/or pcreposix.rc (as is already done for MinGW
947 builds). The .rc files can be used to set FileDescription and many other
948 attributes.
949
950 2. Add an option (-DINSTALL_MSVC_PDB) to enable installation of .pdb files.
951 This allows higher-level build scripts which want .pdb files to avoid
952 hard-coding the exact files needed."
953
954 42. Added support for [[:<:]] and [[:>:]] as used in the BSD POSIX library to
955 mean "start of word" and "end of word", respectively, as a transition aid.
956
957 43. A minimizing repeat of a class containing codepoints greater than 255 in
958 non-UTF 16-bit or 32-bit modes caused an internal error when PCRE was
959 compiled to use the heap for recursion.
960
961 44. Got rid of some compiler warnings for unused variables when UTF but not UCP
962 is configured.
963
964
965 Version 8.33 28-May-2013
966 ------------------------
967
968 1. Added 'U' to some constants that are compared to unsigned integers, to
969 avoid compiler signed/unsigned warnings. Added (int) casts to unsigned
970 variables that are added to signed variables, to ensure the result is
971 signed and can be negated.
972
973 2. Applied patch by Daniel Richard G for quashing MSVC warnings to the
974 CMake config files.
975
976 3. Revise the creation of config.h.generic so that all boolean macros are
977 #undefined, whereas non-boolean macros are #ifndef/#endif-ed. This makes
978 overriding via -D on the command line possible.
979
980 4. Changing the definition of the variable "op" in pcre_exec.c from pcre_uchar
981 to unsigned int is reported to make a quite noticeable speed difference in
982 a specific Windows environment. Testing on Linux did also appear to show
983 some benefit (and it is clearly not harmful). Also fixed the definition of
984 Xop which should be unsigned.
985
986 5. Related to (4), changing the definition of the intermediate variable cc
987 in repeated character loops from pcre_uchar to pcre_uint32 also gave speed
988 improvements.
989
990 6. Fix forward search in JIT when link size is 3 or greater. Also removed some
991 unnecessary spaces.
992
993 7. Adjust autogen.sh and configure.ac to lose warnings given by automake 1.12
994 and later.
995
996 8. Fix two buffer over read issues in 16 and 32 bit modes. Affects JIT only.
997
998 9. Optimizing fast_forward_start_bits in JIT.
999
1000 10. Adding support for callouts in JIT, and fixing some issues revealed
1001 during this work. Namely:
1002
1003 (a) Unoptimized capturing brackets incorrectly reset on backtrack.
1004
1005 (b) Minimum length was not checked before the matching is started.
1006
1007 11. The value of capture_last that is passed to callouts was incorrect in some
1008 cases when there was a capture on one path that was subsequently abandoned
1009 after a backtrack. Also, the capture_last value is now reset after a
1010 recursion, since all captures are also reset in this case.
1011
1012 12. The interpreter no longer returns the "too many substrings" error in the
1013 case when an overflowing capture is in a branch that is subsequently
1014 abandoned after a backtrack.
1015
1016 13. In the pathological case when an offset vector of size 2 is used, pcretest
1017 now prints out the matched string after a yield of 0 or 1.
1018
1019 14. Inlining subpatterns in recursions, when certain conditions are fulfilled.
1020 Only supported by the JIT compiler at the moment.
1021
1022 15. JIT compiler now supports 32 bit Macs thanks to Lawrence Velazquez.
1023
1024 16. Partial matches now set offsets[2] to the "bumpalong" value, that is, the
1025 offset of the starting point of the matching process, provided the offsets
1026 vector is large enough.
1027
1028 17. The \A escape now records a lookbehind value of 1, though its execution
1029 does not actually inspect the previous character. This is to ensure that,
1030 in partial multi-segment matching, at least one character from the old
1031 segment is retained when a new segment is processed. Otherwise, if there
1032 are no lookbehinds in the pattern, \A might match incorrectly at the start
1033 of a new segment.
1034
1035 18. Added some #ifdef __VMS code into pcretest.c to help VMS implementations.
1036
1037 19. Redefined some pcre_uchar variables in pcre_exec.c as pcre_uint32; this
1038 gives some modest performance improvement in 8-bit mode.
1039
1040 20. Added the PCRE-specific property \p{Xuc} for matching characters that can
1041 be expressed in certain programming languages using Universal Character
1042 Names.
1043
1044 21. Unicode validation has been updated in the light of Unicode Corrigendum #9,
1045 which points out that "non characters" are not "characters that may not
1046 appear in Unicode strings" but rather "characters that are reserved for
1047 internal use and have only local meaning".
1048
1049 22. When a pattern was compiled with automatic callouts (PCRE_AUTO_CALLOUT) and
1050 there was a conditional group that depended on an assertion, if the
1051 assertion was false, the callout that immediately followed the alternation
1052 in the condition was skipped when pcre_exec() was used for matching.
1053
1054 23. Allow an explicit callout to be inserted before an assertion that is the
1055 condition for a conditional group, for compatibility with automatic
1056 callouts, which always insert a callout at this point.
1057
1058 24. In 8.31, (*COMMIT) was confined to within a recursive subpattern. Perl also
1059 confines (*SKIP) and (*PRUNE) in the same way, and this has now been done.
1060
1061 25. (*PRUNE) is now supported by the JIT compiler.
1062
1063 26. Fix infinite loop when /(?<=(*SKIP)ac)a/ is matched against aa.
1064
1065 27. Fix the case where there are two or more SKIPs with arguments that may be
1066 ignored.
1067
1068 28. (*SKIP) is now supported by the JIT compiler.
1069
1070 29. (*THEN) is now supported by the JIT compiler.
1071
1072 30. Update RunTest with additional test selector options.
1073
1074 31. The way PCRE handles backtracking verbs has been changed in two ways.
1075
1076 (1) Previously, in something like (*COMMIT)(*SKIP), COMMIT would override
1077 SKIP. Now, PCRE acts on whichever backtracking verb is reached first by
1078 backtracking. In some cases this makes it more Perl-compatible, but Perl's
1079 rather obscure rules do not always do the same thing.
1080
1081 (2) Previously, backtracking verbs were confined within assertions. This is
1082 no longer the case for positive assertions, except for (*ACCEPT). Again,
1083 this sometimes improves Perl compatibility, and sometimes does not.
1084
1085 32. A number of tests that were in test 2 because Perl did things differently
1086 have been moved to test 1, because either Perl or PCRE has changed, and
1087 these tests are now compatible.
1088
1089 32. Backtracking control verbs are now handled in the same way in JIT and
1090 interpreter.
1091
1092 33. An opening parenthesis in a MARK/PRUNE/SKIP/THEN name in a pattern that
1093 contained a forward subroutine reference caused a compile error.
1094
1095 34. Auto-detect and optimize limited repetitions in JIT.
1096
1097 35. Implement PCRE_NEVER_UTF to lock out the use of UTF, in particular,
1098 blocking (*UTF) etc.
1099
1100 36. In the interpreter, maximizing pattern repetitions for characters and
1101 character types now use tail recursion, which reduces stack usage.
1102
1103 37. The value of the max lookbehind was not correctly preserved if a compiled
1104 and saved regex was reloaded on a host of different endianness.
1105
1106 38. Implemented (*LIMIT_MATCH) and (*LIMIT_RECURSION). As part of the extension
1107 of the compiled pattern block, expand the flags field from 16 to 32 bits
1108 because it was almost full.
1109
1110 39. Try madvise first before posix_madvise.
1111
1112 40. Change 7 for PCRE 7.9 made it impossible for pcregrep to find empty lines
1113 with a pattern such as ^$. It has taken 4 years for anybody to notice! The
1114 original change locked out all matches of empty strings. This has been
1115 changed so that one match of an empty string per line is recognized.
1116 Subsequent searches on the same line (for colouring or for --only-matching,
1117 for example) do not recognize empty strings.
1118
1119 41. Applied a user patch to fix a number of spelling mistakes in comments.
1120
1121 42. Data lines longer than 65536 caused pcretest to crash.
1122
1123 43. Clarified the data type for length and startoffset arguments for pcre_exec
1124 and pcre_dfa_exec in the function-specific man pages, where they were
1125 explicitly stated to be in bytes, never having been updated. I also added
1126 some clarification to the pcreapi man page.
1127
1128 44. A call to pcre_dfa_exec() with an output vector size less than 2 caused
1129 a segmentation fault.
1130
1131
1132 Version 8.32 30-November-2012
1133 -----------------------------
1134
1135 1. Improved JIT compiler optimizations for first character search and single
1136 character iterators.
1137
1138 2. Supporting IBM XL C compilers for PPC architectures in the JIT compiler.
1139 Patch by Daniel Richard G.
1140
1141 3. Single character iterator optimizations in the JIT compiler.
1142
1143 4. Improved JIT compiler optimizations for character ranges.
1144
1145 5. Rename the "leave" variable names to "quit" to improve WinCE compatibility.
1146 Reported by Giuseppe D'Angelo.
1147
1148 6. The PCRE_STARTLINE bit, indicating that a match can occur only at the start
1149 of a line, was being set incorrectly in cases where .* appeared inside
1150 atomic brackets at the start of a pattern, or where there was a subsequent
1151 *PRUNE or *SKIP.
1152
1153 7. Improved instruction cache flush for POWER/PowerPC.
1154 Patch by Daniel Richard G.
1155
1156 8. Fixed a number of issues in pcregrep, making it more compatible with GNU
1157 grep:
1158
1159 (a) There is now no limit to the number of patterns to be matched.
1160
1161 (b) An error is given if a pattern is too long.
1162
1163 (c) Multiple uses of --exclude, --exclude-dir, --include, and --include-dir
1164 are now supported.
1165
1166 (d) --exclude-from and --include-from (multiple use) have been added.
1167
1168 (e) Exclusions and inclusions now apply to all files and directories, not
1169 just to those obtained from scanning a directory recursively.
1170
1171 (f) Multiple uses of -f and --file-list are now supported.
1172
1173 (g) In a Windows environment, the default for -d has been changed from
1174 "read" (the GNU grep default) to "skip", because otherwise the presence
1175 of a directory in the file list provokes an error.
1176
1177 (h) The documentation has been revised and clarified in places.
1178
1179 9. Improve the matching speed of capturing brackets.
1180
1181 10. Changed the meaning of \X so that it now matches a Unicode extended
1182 grapheme cluster.
1183
1184 11. Patch by Daniel Richard G to the autoconf files to add a macro for sorting
1185 out POSIX threads when JIT support is configured.
1186
1187 12. Added support for PCRE_STUDY_EXTRA_NEEDED.
1188
1189 13. In the POSIX wrapper regcomp() function, setting re_nsub field in the preg
1190 structure could go wrong in environments where size_t is not the same size
1191 as int.
1192
1193 14. Applied user-supplied patch to pcrecpp.cc to allow PCRE_NO_UTF8_CHECK to be
1194 set.
1195
1196 15. The EBCDIC support had decayed; later updates to the code had included
1197 explicit references to (e.g.) \x0a instead of CHAR_LF. There has been a
1198 general tidy up of EBCDIC-related issues, and the documentation was also
1199 not quite right. There is now a test that can be run on ASCII systems to
1200 check some of the EBCDIC-related things (but is it not a full test).
1201
1202 16. The new PCRE_STUDY_EXTRA_NEEDED option is now used by pcregrep, resulting
1203 in a small tidy to the code.
1204
1205 17. Fix JIT tests when UTF is disabled and both 8 and 16 bit mode are enabled.
1206
1207 18. If the --only-matching (-o) option in pcregrep is specified multiple
1208 times, each one causes appropriate output. For example, -o1 -o2 outputs the
1209 substrings matched by the 1st and 2nd capturing parentheses. A separating
1210 string can be specified by --om-separator (default empty).
1211
1212 19. Improving the first n character searches.
1213
1214 20. Turn case lists for horizontal and vertical white space into macros so that
1215 they are defined only once.
1216
1217 21. This set of changes together give more compatible Unicode case-folding
1218 behaviour for characters that have more than one other case when UCP
1219 support is available.
1220
1221 (a) The Unicode property table now has offsets into a new table of sets of
1222 three or more characters that are case-equivalent. The MultiStage2.py
1223 script that generates these tables (the pcre_ucd.c file) now scans
1224 CaseFolding.txt instead of UnicodeData.txt for character case
1225 information.
1226
1227 (b) The code for adding characters or ranges of characters to a character
1228 class has been abstracted into a generalized function that also handles
1229 case-independence. In UTF-mode with UCP support, this uses the new data
1230 to handle characters with more than one other case.
1231
1232 (c) A bug that is fixed as a result of (b) is that codepoints less than 256
1233 whose other case is greater than 256 are now correctly matched
1234 caselessly. Previously, the high codepoint matched the low one, but not
1235 vice versa.
1236
1237 (d) The processing of \h, \H, \v, and \ in character classes now makes use
1238 of the new class addition function, using character lists defined as
1239 macros alongside the case definitions of 20 above.
1240
1241 (e) Caseless back references now work with characters that have more than
1242 one other case.
1243
1244 (f) General caseless matching of characters with more than one other case
1245 is supported.
1246
1247 22. Unicode character properties were updated from Unicode 6.2.0
1248
1249 23. Improved CMake support under Windows. Patch by Daniel Richard G.
1250
1251 24. Add support for 32-bit character strings, and UTF-32
1252
1253 25. Major JIT compiler update (code refactoring and bugfixing).
1254 Experimental Sparc 32 support is added.
1255
1256 26. Applied a modified version of Daniel Richard G's patch to create
1257 pcre.h.generic and config.h.generic by "make" instead of in the
1258 PrepareRelease script.
1259
1260 27. Added a definition for CHAR_NULL (helpful for the z/OS port), and use it in
1261 pcre_compile.c when checking for a zero character.
1262
1263 28. Introducing a native interface for JIT. Through this interface, the compiled
1264 machine code can be directly executed. The purpose of this interface is to
1265 provide fast pattern matching, so several sanity checks are not performed.
1266 However, feature tests are still performed. The new interface provides
1267 1.4x speedup compared to the old one.
1268
1269 29. If pcre_exec() or pcre_dfa_exec() was called with a negative value for
1270 the subject string length, the error given was PCRE_ERROR_BADOFFSET, which
1271 was confusing. There is now a new error PCRE_ERROR_BADLENGTH for this case.
1272
1273 30. In 8-bit UTF-8 mode, pcretest failed to give an error for data codepoints
1274 greater than 0x7fffffff (which cannot be represented in UTF-8, even under
1275 the "old" RFC 2279). Instead, it ended up passing a negative length to
1276 pcre_exec().
1277
1278 31. Add support for GCC's visibility feature to hide internal functions.
1279
1280 32. Running "pcretest -C pcre8" or "pcretest -C pcre16" gave a spurious error
1281 "unknown -C option" after outputting 0 or 1.
1282
1283 33. There is now support for generating a code coverage report for the test
1284 suite in environments where gcc is the compiler and lcov is installed. This
1285 is mainly for the benefit of the developers.
1286
1287 34. If PCRE is built with --enable-valgrind, certain memory regions are marked
1288 unaddressable using valgrind annotations, allowing valgrind to detect
1289 invalid memory accesses. This is mainly for the benefit of the developers.
1290
1291 25. (*UTF) can now be used to start a pattern in any of the three libraries.
1292
1293 26. Give configure error if --enable-cpp but no C++ compiler found.
1294
1295
1296 Version 8.31 06-July-2012
1297 -------------------------
1298
1299 1. Fixing a wrong JIT test case and some compiler warnings.
1300
1301 2. Removed a bashism from the RunTest script.
1302
1303 3. Add a cast to pcre_exec.c to fix the warning "unary minus operator applied
1304 to unsigned type, result still unsigned" that was given by an MS compiler
1305 on encountering the code "-sizeof(xxx)".
1306
1307 4. Partial matching support is added to the JIT compiler.
1308
1309 5. Fixed several bugs concerned with partial matching of items that consist
1310 of more than one character:
1311
1312 (a) /^(..)\1/ did not partially match "aba" because checking references was
1313 done on an "all or nothing" basis. This also applied to repeated
1314 references.
1315
1316 (b) \R did not give a hard partial match if \r was found at the end of the
1317 subject.
1318
1319 (c) \X did not give a hard partial match after matching one or more
1320 characters at the end of the subject.
1321
1322 (d) When newline was set to CRLF, a pattern such as /a$/ did not recognize
1323 a partial match for the string "\r".
1324
1325 (e) When newline was set to CRLF, the metacharacter "." did not recognize
1326 a partial match for a CR character at the end of the subject string.
1327
1328 6. If JIT is requested using /S++ or -s++ (instead of just /S+ or -s+) when
1329 running pcretest, the text "(JIT)" added to the output whenever JIT is
1330 actually used to run the match.
1331
1332 7. Individual JIT compile options can be set in pcretest by following -s+[+]
1333 or /S+[+] with a digit between 1 and 7.
1334
1335 8. OP_NOT now supports any UTF character not just single-byte ones.
1336
1337 9. (*MARK) control verb is now supported by the JIT compiler.
1338
1339 10. The command "./RunTest list" lists the available tests without actually
1340 running any of them. (Because I keep forgetting what they all are.)
1341
1342 11. Add PCRE_INFO_MAXLOOKBEHIND.
1343
1344 12. Applied a (slightly modified) user-supplied patch that improves performance
1345 when the heap is used for recursion (compiled with --disable-stack-for-
1346 recursion). Instead of malloc and free for each heap frame each time a
1347 logical recursion happens, frames are retained on a chain and re-used where
1348 possible. This sometimes gives as much as 30% improvement.
1349
1350 13. As documented, (*COMMIT) is now confined to within a recursive subpattern
1351 call.
1352
1353 14. As documented, (*COMMIT) is now confined to within a positive assertion.
1354
1355 15. It is now possible to link pcretest with libedit as an alternative to
1356 libreadline.
1357
1358 16. (*COMMIT) control verb is now supported by the JIT compiler.
1359
1360 17. The Unicode data tables have been updated to Unicode 6.1.0.
1361
1362 18. Added --file-list option to pcregrep.
1363
1364 19. Added binary file support to pcregrep, including the -a, --binary-files,
1365 -I, and --text options.
1366
1367 20. The madvise function is renamed for posix_madvise for QNX compatibility
1368 reasons. Fixed by Giuseppe D'Angelo.
1369
1370 21. Fixed a bug for backward assertions with REVERSE 0 in the JIT compiler.
1371
1372 22. Changed the option for creating symbolic links for 16-bit man pages from
1373 -s to -sf so that re-installing does not cause issues.
1374
1375 23. Support PCRE_NO_START_OPTIMIZE in JIT as (*MARK) support requires it.
1376
1377 24. Fixed a very old bug in pcretest that caused errors with restarted DFA
1378 matches in certain environments (the workspace was not being correctly
1379 retained). Also added to pcre_dfa_exec() a simple plausibility check on
1380 some of the workspace data at the beginning of a restart.
1381
1382 25. \s*\R was auto-possessifying the \s* when it should not, whereas \S*\R
1383 was not doing so when it should - probably a typo introduced by SVN 528
1384 (change 8.10/14).
1385
1386 26. When PCRE_UCP was not set, \w+\x{c4} was incorrectly auto-possessifying the
1387 \w+ when the character tables indicated that \x{c4} was a word character.
1388 There were several related cases, all because the tests for doing a table
1389 lookup were testing for characters less than 127 instead of 255.
1390
1391 27. If a pattern contains capturing parentheses that are not used in a match,
1392 their slots in the ovector are set to -1. For those that are higher than
1393 any matched groups, this happens at the end of processing. In the case when
1394 there were back references that the ovector was too small to contain
1395 (causing temporary malloc'd memory to be used during matching), and the
1396 highest capturing number was not used, memory off the end of the ovector
1397 was incorrectly being set to -1. (It was using the size of the temporary
1398 memory instead of the true size.)
1399
1400 28. To catch bugs like 27 using valgrind, when pcretest is asked to specify an
1401 ovector size, it uses memory at the end of the block that it has got.
1402
1403 29. Check for an overlong MARK name and give an error at compile time. The
1404 limit is 255 for the 8-bit library and 65535 for the 16-bit library.
1405
1406 30. JIT compiler update.
1407
1408 31. JIT is now supported on jailbroken iOS devices. Thanks for Ruiger
1409 Rill for the patch.
1410
1411 32. Put spaces around SLJIT_PRINT_D in the JIT compiler. Required by CXX11.
1412
1413 33. Variable renamings in the PCRE-JIT compiler. No functionality change.
1414
1415 34. Fixed typos in pcregrep: in two places there was SUPPORT_LIBZ2 instead of
1416 SUPPORT_LIBBZ2. This caused a build problem when bzip2 but not gzip (zlib)
1417 was enabled.
1418
1419 35. Improve JIT code generation for greedy plus quantifier.
1420
1421 36. When /((?:a?)*)*c/ or /((?>a?)*)*c/ was matched against "aac", it set group
1422 1 to "aa" instead of to an empty string. The bug affected repeated groups
1423 that could potentially match an empty string.
1424
1425 37. Optimizing single character iterators in JIT.
1426
1427 38. Wide characters specified with \uxxxx in JavaScript mode are now subject to
1428 the same checks as \x{...} characters in non-JavaScript mode. Specifically,
1429 codepoints that are too big for the mode are faulted, and in a UTF mode,
1430 disallowed codepoints are also faulted.
1431
1432 39. If PCRE was compiled with UTF support, in three places in the DFA
1433 matcher there was code that should only have been obeyed in UTF mode, but
1434 was being obeyed unconditionally. In 8-bit mode this could cause incorrect
1435 processing when bytes with values greater than 127 were present. In 16-bit
1436 mode the bug would be provoked by values in the range 0xfc00 to 0xdc00. In
1437 both cases the values are those that cannot be the first data item in a UTF
1438 character. The three items that might have provoked this were recursions,
1439 possessively repeated groups, and atomic groups.
1440
1441 40. Ensure that libpcre is explicitly listed in the link commands for pcretest
1442 and pcregrep, because some OS require shared objects to be explicitly
1443 passed to ld, causing the link step to fail if they are not.
1444
1445 41. There were two incorrect #ifdefs in pcre_study.c, meaning that, in 16-bit
1446 mode, patterns that started with \h* or \R* might be incorrectly matched.
1447
1448
1449 Version 8.30 04-February-2012
1450 -----------------------------
1451
1452 1. Renamed "isnumber" as "is_a_number" because in some Mac environments this
1453 name is defined in ctype.h.
1454
1455 2. Fixed a bug in fixed-length calculation for lookbehinds that would show up
1456 only in quite long subpatterns.
1457
1458 3. Removed the function pcre_info(), which has been obsolete and deprecated
1459 since it was replaced by pcre_fullinfo() in February 2000.
1460
1461 4. For a non-anchored pattern, if (*SKIP) was given with a name that did not
1462 match a (*MARK), and the match failed at the start of the subject, a
1463 reference to memory before the start of the subject could occur. This bug
1464 was introduced by fix 17 of release 8.21.
1465
1466 5. A reference to an unset group with zero minimum repetition was giving
1467 totally wrong answers (in non-JavaScript-compatibility mode). For example,
1468 /(another)?(\1?)test/ matched against "hello world test". This bug was
1469 introduced in release 8.13.
1470
1471 6. Add support for 16-bit character strings (a large amount of work involving
1472 many changes and refactorings).
1473
1474 7. RunGrepTest failed on msys because \r\n was replaced by whitespace when the
1475 command "pattern=`printf 'xxx\r\njkl'`" was run. The pattern is now taken
1476 from a file.
1477
1478 8. Ovector size of 2 is also supported by JIT based pcre_exec (the ovector size
1479 rounding is not applied in this particular case).
1480
1481 9. The invalid Unicode surrogate codepoints U+D800 to U+DFFF are now rejected
1482 if they appear, or are escaped, in patterns.
1483
1484 10. Get rid of a number of -Wunused-but-set-variable warnings.
1485
1486 11. The pattern /(?=(*:x))(q|)/ matches an empty string, and returns the mark
1487 "x". The similar pattern /(?=(*:x))((*:y)q|)/ did not return a mark at all.
1488 Oddly, Perl behaves the same way. PCRE has been fixed so that this pattern
1489 also returns the mark "x". This bug applied to capturing parentheses,
1490 non-capturing parentheses, and atomic parentheses. It also applied to some
1491 assertions.
1492
1493 12. Stephen Kelly's patch to CMakeLists.txt allows it to parse the version
1494 information out of configure.ac instead of relying on pcre.h.generic, which
1495 is not stored in the repository.
1496
1497 13. Applied Dmitry V. Levin's patch for a more portable method for linking with
1498 -lreadline.
1499
1500 14. ZH added PCRE_CONFIG_JITTARGET; added its output to pcretest -C.
1501
1502 15. Applied Graycode's patch to put the top-level frame on the stack rather
1503 than the heap when not using the stack for recursion. This gives a
1504 performance improvement in many cases when recursion is not deep.
1505
1506 16. Experimental code added to "pcretest -C" to output the stack frame size.
1507
1508
1509 Version 8.21 12-Dec-2011
1510 ------------------------
1511
1512 1. Updating the JIT compiler.
1513
1514 2. JIT compiler now supports OP_NCREF, OP_RREF and OP_NRREF. New test cases
1515 are added as well.
1516
1517 3. Fix cache-flush issue on PowerPC (It is still an experimental JIT port).
1518 PCRE_EXTRA_TABLES is not suported by JIT, and should be checked before
1519 calling _pcre_jit_exec. Some extra comments are added.
1520
1521 4. (*MARK) settings inside atomic groups that do not contain any capturing
1522 parentheses, for example, (?>a(*:m)), were not being passed out. This bug
1523 was introduced by change 18 for 8.20.
1524
1525 5. Supporting of \x, \U and \u in JavaScript compatibility mode based on the
1526 ECMA-262 standard.
1527
1528 6. Lookbehinds such as (?<=a{2}b) that contained a fixed repetition were
1529 erroneously being rejected as "not fixed length" if PCRE_CASELESS was set.
1530 This bug was probably introduced by change 9 of 8.13.
1531
1532 7. While fixing 6 above, I noticed that a number of other items were being
1533 incorrectly rejected as "not fixed length". This arose partly because newer
1534 opcodes had not been added to the fixed-length checking code. I have (a)
1535 corrected the bug and added tests for these items, and (b) arranged for an
1536 error to occur if an unknown opcode is encountered while checking for fixed
1537 length instead of just assuming "not fixed length". The items that were
1538 rejected were: (*ACCEPT), (*COMMIT), (*FAIL), (*MARK), (*PRUNE), (*SKIP),
1539 (*THEN), \h, \H, \v, \V, and single character negative classes with fixed
1540 repetitions, e.g. [^a]{3}, with and without PCRE_CASELESS.
1541
1542 8. A possessively repeated conditional subpattern such as (?(?=c)c|d)++ was
1543 being incorrectly compiled and would have given unpredicatble results.
1544
1545 9. A possessively repeated subpattern with minimum repeat count greater than
1546 one behaved incorrectly. For example, (A){2,}+ behaved as if it was
1547 (A)(A)++ which meant that, after a subsequent mismatch, backtracking into
1548 the first (A) could occur when it should not.
1549
1550 10. Add a cast and remove a redundant test from the code.
1551
1552 11. JIT should use pcre_malloc/pcre_free for allocation.
1553
1554 12. Updated pcre-config so that it no longer shows -L/usr/lib, which seems
1555 best practice nowadays, and helps with cross-compiling. (If the exec_prefix
1556 is anything other than /usr, -L is still shown).
1557
1558 13. In non-UTF-8 mode, \C is now supported in lookbehinds and DFA matching.
1559
1560 14. Perl does not support \N without a following name in a [] class; PCRE now
1561 also gives an error.
1562
1563 15. If a forward reference was repeated with an upper limit of around 2000,
1564 it caused the error "internal error: overran compiling workspace". The
1565 maximum number of forward references (including repeats) was limited by the
1566 internal workspace, and dependent on the LINK_SIZE. The code has been
1567 rewritten so that the workspace expands (via pcre_malloc) if necessary, and
1568 the default depends on LINK_SIZE. There is a new upper limit (for safety)
1569 of around 200,000 forward references. While doing this, I also speeded up
1570 the filling in of repeated forward references.
1571
1572 16. A repeated forward reference in a pattern such as (a)(?2){2}(.) was
1573 incorrectly expecting the subject to contain another "a" after the start.
1574
1575 17. When (*SKIP:name) is activated without a corresponding (*MARK:name) earlier
1576 in the match, the SKIP should be ignored. This was not happening; instead
1577 the SKIP was being treated as NOMATCH. For patterns such as
1578 /A(*MARK:A)A+(*SKIP:B)Z|AAC/ this meant that the AAC branch was never
1579 tested.
1580
1581 18. The behaviour of (*MARK), (*PRUNE), and (*THEN) has been reworked and is
1582 now much more compatible with Perl, in particular in cases where the result
1583 is a non-match for a non-anchored pattern. For example, if
1584 /b(*:m)f|a(*:n)w/ is matched against "abc", the non-match returns the name
1585 "m", where previously it did not return a name. A side effect of this
1586 change is that for partial matches, the last encountered mark name is
1587 returned, as for non matches. A number of tests that were previously not
1588 Perl-compatible have been moved into the Perl-compatible test files. The
1589 refactoring has had the pleasing side effect of removing one argument from
1590 the match() function, thus reducing its stack requirements.
1591
1592 19. If the /S+ option was used in pcretest to study a pattern using JIT,
1593 subsequent uses of /S (without +) incorrectly behaved like /S+.
1594
1595 21. Retrieve executable code size support for the JIT compiler and fixing
1596 some warnings.
1597
1598 22. A caseless match of a UTF-8 character whose other case uses fewer bytes did
1599 not work when the shorter character appeared right at the end of the
1600 subject string.
1601
1602 23. Added some (int) casts to non-JIT modules to reduce warnings on 64-bit
1603 systems.
1604
1605 24. Added PCRE_INFO_JITSIZE to pass on the value from (21) above, and also
1606 output it when the /M option is used in pcretest.
1607
1608 25. The CheckMan script was not being included in the distribution. Also, added
1609 an explicit "perl" to run Perl scripts from the PrepareRelease script
1610 because this is reportedly needed in Windows.
1611
1612 26. If study data was being save in a file and studying had not found a set of
1613 "starts with" bytes for the pattern, the data written to the file (though
1614 never used) was taken from uninitialized memory and so caused valgrind to
1615 complain.
1616
1617 27. Updated RunTest.bat as provided by Sheri Pierce.
1618
1619 28. Fixed a possible uninitialized memory bug in pcre_jit_compile.c.
1620
1621 29. Computation of memory usage for the table of capturing group names was
1622 giving an unnecessarily large value.
1623
1624
1625 Version 8.20 21-Oct-2011
1626 ------------------------
1627
1628 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
1629 a POSIX class. After further experiments with Perl, which convinced me that
1630 Perl has bugs and confusions, a closing square bracket is no longer allowed
1631 in a POSIX name. This bug also affected patterns with classes that started
1632 with full stops.
1633
1634 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no
1635 captured substring, but while checking the failing first alternative,
1636 substring 1 is temporarily captured. If the output vector supplied to
1637 pcre_exec() was not big enough for this capture, the yield of the function
1638 was still zero ("insufficient space for captured substrings"). This cannot
1639 be totally fixed without adding another stack variable, which seems a lot
1640 of expense for a edge case. However, I have improved the situation in cases
1641 such as /(a)(b)x|abc/ matched against "abc", where the return code
1642 indicates that fewer than the maximum number of slots in the ovector have
1643 been set.
1644
1645 3. Related to (2) above: when there are more back references in a pattern than
1646 slots in the output vector, pcre_exec() uses temporary memory during
1647 matching, and copies in the captures as far as possible afterwards. It was
1648 using the entire output vector, but this conflicts with the specification
1649 that only 2/3 is used for passing back captured substrings. Now it uses
1650 only the first 2/3, for compatibility. This is, of course, another edge
1651 case.
1652
1653 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
1654 main code base, and can be used by building with --enable-jit. When this is
1655 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
1656 runtime --no-jit option is given.
1657
1658 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
1659 ovector, the return from the function was zero, implying that there were
1660 other matches that did not fit. The correct "exactly full" value is now
1661 returned.
1662
1663 6. If a subpattern that was called recursively or as a subroutine contained
1664 (*PRUNE) or any other control that caused it to give a non-standard return,
1665 invalid errors such as "Error -26 (nested recursion at the same subject
1666 position)" or even infinite loops could occur.
1667
1668 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
1669 computing the minimum length on reaching *ACCEPT, and so ended up with the
1670 wrong value of 1 rather than 0. Further investigation indicates that
1671 computing a minimum subject length in the presence of *ACCEPT is difficult
1672 (think back references, subroutine calls), and so I have changed the code
1673 so that no minimum is registered for a pattern that contains *ACCEPT.
1674
1675 8. If (*THEN) was present in the first (true) branch of a conditional group,
1676 it was not handled as intended. [But see 16 below.]
1677
1678 9. Replaced RunTest.bat and CMakeLists.txt with improved versions provided by
1679 Sheri Pierce.
1680
1681 10. A pathological pattern such as /(*ACCEPT)a/ was miscompiled, thinking that
1682 the first byte in a match must be "a".
1683
1684 11. Change 17 for 8.13 increased the recursion depth for patterns like
1685 /a(?:.)*?a/ drastically. I've improved things by remembering whether a
1686 pattern contains any instances of (*THEN). If it does not, the old
1687 optimizations are restored. It would be nice to do this on a per-group
1688 basis, but at the moment that is not feasible.
1689
1690 12. In some environments, the output of pcretest -C is CRLF terminated. This
1691 broke RunTest's code that checks for the link size. A single white space
1692 character after the value is now allowed for.
1693
1694 13. RunTest now checks for the "fr" locale as well as for "fr_FR" and "french".
1695 For "fr", it uses the Windows-specific input and output files.
1696
1697 14. If (*THEN) appeared in a group that was called recursively or as a
1698 subroutine, it did not work as intended. [But see next item.]
1699
1700 15. Consider the pattern /A (B(*THEN)C) | D/ where A, B, C, and D are complex
1701 pattern fragments (but not containing any | characters). If A and B are
1702 matched, but there is a failure in C so that it backtracks to (*THEN), PCRE
1703 was behaving differently to Perl. PCRE backtracked into A, but Perl goes to
1704 D. In other words, Perl considers parentheses that do not contain any |
1705 characters to be part of a surrounding alternative, whereas PCRE was
1706 treading (B(*THEN)C) the same as (B(*THEN)C|(*FAIL)) -- which Perl handles
1707 differently. PCRE now behaves in the same way as Perl, except in the case
1708 of subroutine/recursion calls such as (?1) which have in any case always
1709 been different (but PCRE had them first :-).
1710
1711 16. Related to 15 above: Perl does not treat the | in a conditional group as
1712 creating alternatives. Such a group is treated in the same way as an
1713 ordinary group without any | characters when processing (*THEN). PCRE has
1714 been changed to match Perl's behaviour.
1715
1716 17. If a user had set PCREGREP_COLO(U)R to something other than 1:31, the
1717 RunGrepTest script failed.
1718
1719 18. Change 22 for version 13 caused atomic groups to use more stack. This is
1720 inevitable for groups that contain captures, but it can lead to a lot of
1721 stack use in large patterns. The old behaviour has been restored for atomic
1722 groups that do not contain any capturing parentheses.
1723
1724 19. If the PCRE_NO_START_OPTIMIZE option was set for pcre_compile(), it did not
1725 suppress the check for a minimum subject length at run time. (If it was
1726 given to pcre_exec() or pcre_dfa_exec() it did work.)
1727
1728 20. Fixed an ASCII-dependent infelicity in pcretest that would have made it
1729 fail to work when decoding hex characters in data strings in EBCDIC
1730 environments.
1731
1732 21. It appears that in at least one Mac OS environment, the isxdigit() function
1733 is implemented as a macro that evaluates to its argument more than once,
1734 contravening the C 90 Standard (I haven't checked a later standard). There
1735 was an instance in pcretest which caused it to go wrong when processing
1736 \x{...} escapes in subject strings. The has been rewritten to avoid using
1737 things like p++ in the argument of isxdigit().
1738
1739
1740 Version 8.13 16-Aug-2011
1741 ------------------------
1742
1743 1. The Unicode data tables have been updated to Unicode 6.0.0.
1744
1745 2. Two minor typos in pcre_internal.h have been fixed.
1746
1747 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
1748 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
1749 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
1750
1751 4. There were a number of related bugs in the code for matching backrefences
1752 caselessly in UTF-8 mode when codes for the characters concerned were
1753 different numbers of bytes. For example, U+023A and U+2C65 are an upper
1754 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
1755 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
1756 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
1757 2-byte code at the end of the subject (it thought there wasn't enough data
1758 left).
1759
1760 5. Comprehensive information about what went wrong is now returned by
1761 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
1762 as the output vector has at least 2 elements. The offset of the start of
1763 the failing character and a reason code are placed in the vector.
1764
1765 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
1766 now returned is for the first byte of the failing character, instead of the
1767 last byte inspected. This is an incompatible change, but I hope it is small
1768 enough not to be a problem. It makes the returned offset consistent with
1769 pcre_exec() and pcre_dfa_exec().
1770
1771 7. pcretest now gives a text phrase as well as the error number when
1772 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
1773 failure, the offset and reason code are output.
1774
1775 8. When \R was used with a maximizing quantifier it failed to skip backwards
1776 over a \r\n pair if the subsequent match failed. Instead, it just skipped
1777 back over a single character (\n). This seems wrong (because it treated the
1778 two characters as a single entity when going forwards), conflicts with the
1779 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
1780 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
1781 has been changed.
1782
1783 9. Some internal refactoring has changed the processing so that the handling
1784 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
1785 time (the PCRE_DOTALL option was changed this way some time ago: version
1786 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
1787 which was always a bit of a fudge. It also means that there is one less
1788 argument for the match() function, which reduces its stack requirements
1789 slightly. This change also fixes an incompatibility with Perl: the pattern
1790 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
1791
1792 10. More internal refactoring has drastically reduced the number of recursive
1793 calls to match() for possessively repeated groups such as (abc)++ when
1794 using pcre_exec().
1795
1796 11. While implementing 10, a number of bugs in the handling of groups were
1797 discovered and fixed:
1798
1799 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
1800 (a|)*(?1) gave a compile-time internal error.
1801 ((a|)+)+ did not notice that the outer group could match an empty string.
1802 (^a|^)+ was not marked as anchored.
1803 (.*a|.*)+ was not marked as matching at start or after a newline.
1804
1805 12. Yet more internal refactoring has removed another argument from the match()
1806 function. Special calls to this function are now indicated by setting a
1807 value in a variable in the "match data" data block.
1808
1809 13. Be more explicit in pcre_study() instead of relying on "default" for
1810 opcodes that mean there is no starting character; this means that when new
1811 ones are added and accidentally left out of pcre_study(), testing should
1812 pick them up.
1813
1814 14. The -s option of pcretest has been documented for ages as being an old
1815 synonym of -m (show memory usage). I have changed it to mean "force study
1816 for every regex", that is, assume /S for every regex. This is similar to -i
1817 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
1818 using it. It makes it easier to run collections of tests with and without
1819 study enabled, and thereby test pcre_study() more easily. All the standard
1820 tests are now run with and without -s (but some patterns can be marked as
1821 "never study" - see 20 below).
1822
1823 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
1824 restoration of the capturing data to the outer values was not happening
1825 correctly.
1826
1827 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
1828 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
1829 pattern had matched an empty string, and so incorrectly returned a no
1830 match.
1831
1832 17. There was optimizing code for the last branch of non-capturing parentheses,
1833 and also for the obeyed branch of a conditional subexpression, which used
1834 tail recursion to cut down on stack usage. Unfortunately, now that there is
1835 the possibility of (*THEN) occurring in these branches, tail recursion is
1836 no longer possible because the return has to be checked for (*THEN). These
1837 two optimizations have therefore been removed. [But see 8.20/11 above.]
1838
1839 18. If a pattern containing \R was studied, it was assumed that \R always
1840 matched two bytes, thus causing the minimum subject length to be
1841 incorrectly computed because \R can also match just one byte.
1842
1843 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
1844 was incorrectly computed.
1845
1846 20. If /S is present twice on a test pattern in pcretest input, it now
1847 *disables* studying, thereby overriding the use of -s on the command line
1848 (see 14 above). This is necessary for one or two tests to keep the output
1849 identical in both cases.
1850
1851 21. When (*ACCEPT) was used in an assertion that matched an empty string and
1852 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
1853
1854 22. When an atomic group that contained a capturing parenthesis was
1855 successfully matched, but the branch in which it appeared failed, the
1856 capturing was not being forgotten if a higher numbered group was later
1857 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
1858 group 1 to "a", when in fact it should be unset. This applied to multi-
1859 branched capturing and non-capturing groups, repeated or not, and also to
1860 positive assertions (capturing in negative assertions does not happen
1861 in PCRE) and also to nested atomic groups.
1862
1863 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
1864 subject after a captured substring, to make it easier to tell which of a
1865 number of identical substrings has been captured.
1866
1867 24. The way atomic groups are processed by pcre_exec() has been changed so that
1868 if they are repeated, backtracking one repetition now resets captured
1869 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
1870 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
1871 "aaa". Previously, it would have been "a". As part of this code
1872 refactoring, the way recursive calls are handled has also been changed.
1873
1874 25. If an assertion condition captured any substrings, they were not passed
1875 back unless some other capturing happened later. For example, if
1876 (?(?=(a))a) was matched against "a", no capturing was returned.
1877
1878 26. When studying a pattern that contained subroutine calls or assertions,
1879 the code for finding the minimum length of a possible match was handling
1880 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
1881 group 1 called group 2 while simultaneously a separate group 2 called group
1882 1). A stack overflow occurred in this case. I have fixed this by limiting
1883 the recursion depth to 10.
1884
1885 27. Updated RunTest.bat in the distribution to the version supplied by Tom
1886 Fortmann. This supports explicit test numbers on the command line, and has
1887 argument validation and error reporting.
1888
1889 28. An instance of \X with an unlimited repeat could fail if at any point the
1890 first character it looked at was a mark character.
1891
1892 29. Some minor code refactoring concerning Unicode properties and scripts
1893 should reduce the stack requirement of match() slightly.
1894
1895 30. Added the '=' option to pcretest to check the setting of unused capturing
1896 slots at the end of the pattern, which are documented as being -1, but are
1897 not included in the return count.
1898
1899 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
1900 compiled something random. Now it gives a compile-time error (as does
1901 Perl).
1902
1903 32. A *MARK encountered during the processing of a positive assertion is now
1904 recorded and passed back (compatible with Perl).
1905
1906 33. If --only-matching or --colour was set on a pcregrep call whose pattern
1907 had alternative anchored branches, the search for a second match in a line
1908 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
1909 matched the line "0102" twice. The same bug affected patterns that started
1910 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
1911 twice.
1912
1913 34. Previously, PCRE did not allow quantification of assertions. However, Perl
1914 does, and because of capturing effects, quantifying parenthesized
1915 assertions may at times be useful. Quantifiers are now allowed for
1916 parenthesized assertions.
1917
1918 35. A minor code tidy in pcre_compile() when checking options for \R usage.
1919
1920 36. \g was being checked for fancy things in a character class, when it should
1921 just be a literal "g".
1922
1923 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
1924 appearance of a nested POSIX class supersedes an apparent external class.
1925 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
1926 unescaped square brackets may also appear as part of class names. For
1927 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
1928 more like Perl. (But see 8.20/1 above.)
1929
1930 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
1931 was because it thought it was \N{name}, which is not supported).
1932
1933 39. Add minix to OS list not supporting the -S option in pcretest.
1934
1935 40. PCRE tries to detect cases of infinite recursion at compile time, but it
1936 cannot analyze patterns in sufficient detail to catch mutual recursions
1937 such as ((?1))((?2)). There is now a runtime test that gives an error if a
1938 subgroup is called recursively as a subpattern for a second time at the
1939 same position in the subject string. In previous releases this might have
1940 been caught by the recursion limit, or it might have run out of stack.
1941
1942 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
1943 happen only once. PCRE was, however incorrectly giving a compile time error
1944 "recursive call could loop indefinitely" because it cannot analyze the
1945 pattern in sufficient detail. The compile time test no longer happens when
1946 PCRE is compiling a conditional subpattern, but actual runaway loops are
1947 now caught at runtime (see 40 above).
1948
1949 42. It seems that Perl allows any characters other than a closing parenthesis
1950 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
1951 has been changed to be the same.
1952
1953 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
1954 as not to get warnings when autogen.sh is called. Also changed
1955 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
1956
1957 44. To help people who use pcregrep to scan files containing exceedingly long
1958 lines, the following changes have been made:
1959
1960 (a) The default value of the buffer size parameter has been increased from
1961 8K to 20K. (The actual buffer used is three times this size.)
1962
1963 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
1964 PCRE is built.
1965
1966 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
1967 to be set at run time.
1968
1969 (d) Numerical values in pcregrep options can be followed by K or M, for
1970 example --buffer-size=50K.
1971
1972 (e) If a line being scanned overflows pcregrep's buffer, an error is now
1973 given and the return code is set to 2.
1974
1975 45. Add a pointer to the latest mark to the callout data block.
1976
1977 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
1978 partial match of an empty string instead of no match. This was specific to
1979 the use of ".".
1980
1981 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
1982 complete match instead of a partial match. This bug was dependent on both
1983 the PCRE_UTF8 and PCRE_DOTALL options being set.
1984
1985 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
1986 starting byte set, because \b was not being ignored.
1987
1988
1989 Version 8.12 15-Jan-2011
1990 ------------------------
1991
1992 1. Fixed some typos in the markup of the man pages, and wrote a script that
1993 checks for such things as part of the documentation building process.
1994
1995 2. On a big-endian 64-bit system, pcregrep did not correctly process the
1996 --match-limit and --recursion-limit options (added for 8.11). In
1997 particular, this made one of the standard tests fail. (The integer value
1998 went into the wrong half of a long int.)
1999
2000 3. If the --colour option was given to pcregrep with -v (invert match), it
2001 did strange things, either producing crazy output, or crashing. It should,
2002 of course, ignore a request for colour when reporting lines that do not
2003 match.
2004
2005 4. Another pcregrep bug caused similar problems if --colour was specified with
2006 -M (multiline) and the pattern match finished with a line ending.
2007
2008 5. In pcregrep, when a pattern that ended with a literal newline sequence was
2009 matched in multiline mode, the following line was shown as part of the
2010 match. This seems wrong, so I have changed it.
2011
2012 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
2013 the check for further matches in the same line (so they could be coloured)
2014 to overrun the end of the current line. If another match was found, it was
2015 incorrectly shown (and then shown again when found in the next line).
2016
2017 7. If pcregrep was compiled under Windows, there was a reference to the
2018 function pcregrep_exit() before it was defined. I am assuming this was
2019 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
2020 reported by a user. I've moved the definition above the reference.
2021
2022
2023 Version 8.11 10-Dec-2010
2024 ------------------------
2025
2026 1. (*THEN) was not working properly if there were untried alternatives prior
2027 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
2028 backtracked to try for "b" instead of moving to the next alternative branch
2029 at the same level (in this case, to look for "c"). The Perl documentation
2030 is clear that when (*THEN) is backtracked onto, it goes to the "next
2031 alternative in the innermost enclosing group".
2032
2033 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
2034 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
2035 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
2036 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
2037 (*THEN).
2038
2039 3. If \s appeared in a character class, it removed the VT character from
2040 the class, even if it had been included by some previous item, for example
2041 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
2042 of \s, but is part of the POSIX "space" class.)
2043
2044 4. A partial match never returns an empty string (because you can always
2045 match an empty string at the end of the subject); however the checking for
2046 an empty string was starting at the "start of match" point. This has been
2047 changed to the "earliest inspected character" point, because the returned
2048 data for a partial match starts at this character. This means that, for
2049 example, /(?<=abc)def/ gives a partial match for the subject "abc"
2050 (previously it gave "no match").
2051
2052 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
2053 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
2054 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
2055 has an implication that the given string is incomplete (because a partial
2056 match is preferred over a full match). For this reason, these items now
2057 give a partial match in this situation. [Aside: previously, the one case
2058 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
2059 match rather than a full match, which was wrong by the old rules, but is
2060 now correct.]
2061
2062 6. There was a bug in the handling of #-introduced comments, recognized when
2063 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
2064 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
2065 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
2066 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
2067 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
2068 places in pcre_compile().
2069
2070 7. Related to (6) above, when pcre_compile() was skipping #-introduced
2071 comments when looking ahead for named forward references to subpatterns,
2072 the only newline sequence it recognized was NL. It now handles newlines
2073 according to the set newline convention.
2074
2075 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
2076 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
2077 cater for a lack of strerror(). These oversights have been fixed.
2078
2079 9. Added --match-limit and --recursion-limit to pcregrep.
2080
2081 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
2082
2083 11. When the -o option was used, pcregrep was setting a return code of 1, even
2084 when matches were found, and --line-buffered was not being honoured.
2085
2086 12. Added an optional parentheses number to the -o and --only-matching options
2087 of pcregrep.
2088
2089 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
2090 can match an empty string. The code to do it in pcretest and pcredemo
2091 needed fixing:
2092
2093 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
2094 only one byte after an empty string match just before CRLF (this case
2095 just got forgotten; "any" and "anycrlf" were OK).
2096
2097 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
2098 mode when an empty string match preceded an ASCII character followed by
2099 a non-ASCII character. (The code for advancing by one character rather
2100 than one byte was nonsense.)
2101
2102 (c) The pcredemo.c sample program did not have any code at all to handle
2103 the cases when CRLF is a valid newline sequence.
2104
2105 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
2106 as a starting offset was within the subject string. There is now a new
2107 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
2108 negative or greater than the length of the string. In order to test this,
2109 pcretest is extended to allow the setting of negative starting offsets.
2110
2111 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
2112 starting offset points to the beginning of a UTF-8 character was
2113 unnecessarily clumsy. I tidied it up.
2114
2115 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
2116 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
2117
2118 17. Nobody had reported that the --include_dir option, which was added in
2119 release 7.7 should have been called --include-dir (hyphen, not underscore)
2120 for compatibility with GNU grep. I have changed it to --include-dir, but
2121 left --include_dir as an undocumented synonym, and the same for
2122 --exclude-dir, though that is not available in GNU grep, at least as of
2123 release 2.5.4.
2124
2125 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
2126 characters from a string of bytes) have been redefined so as not to use
2127 loops, in order to improve performance in some environments. At the same
2128 time, I abstracted some of the common code into auxiliary macros to save
2129 repetition (this should not affect the compiled code).
2130
2131 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
2132 compile-time error is now given if \c is not followed by an ASCII
2133 character, that is, a byte less than 128. (In EBCDIC mode, the code is
2134 different, and any byte value is allowed.)
2135
2136 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
2137 START_OPTIMIZE option, which is now allowed at compile time - but just
2138 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
2139 to pcregrep and other applications that have no direct access to PCRE
2140 options. The new /Y option in pcretest sets this option when calling
2141 pcre_compile().
2142
2143 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
2144 back references. Groups containing recursive back references were forced to
2145 be atomic by that change, but in the case of named groups, the amount of
2146 memory required was incorrectly computed, leading to "Failed: internal
2147 error: code overflow". This has been fixed.
2148
2149 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
2150 pcretest.c, to avoid build problems in some Borland environments.
2151
2152
2153 Version 8.10 25-Jun-2010
2154 ------------------------
2155
2156 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
2157 THEN.
2158
2159 2. (*ACCEPT) was not working when inside an atomic group.
2160
2161 3. Inside a character class, \B is treated as a literal by default, but
2162 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
2163 causes the error). The code is unchanged, but I tidied the documentation.
2164
2165 4. Inside a character class, PCRE always treated \R and \X as literals,
2166 whereas Perl faults them if its -w option is set. I have changed PCRE so
2167 that it faults them when PCRE_EXTRA is set.
2168
2169 5. Added support for \N, which always matches any character other than
2170 newline. (It is the same as "." when PCRE_DOTALL is not set.)
2171
2172 6. When compiling pcregrep with newer versions of gcc which may have
2173 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
2174 declared with attribute warn_unused_result" were given. Just casting the
2175 result to (void) does not stop the warnings; a more elaborate fudge is
2176 needed. I've used a macro to implement this.
2177
2178 7. Minor change to pcretest.c to avoid a compiler warning.
2179
2180 8. Added four artifical Unicode properties to help with an option to make
2181 \s etc use properties (see next item). The new properties are: Xan
2182 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
2183
2184 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
2185 use Unicode properties. (*UCP) at the start of a pattern can be used to set
2186 this option. Modified pcretest to add /W to test this facility. Added
2187 REG_UCP to make it available via the POSIX interface.
2188
2189 10. Added --line-buffered to pcregrep.
2190
2191 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
2192 studied, and the match started with a letter with a code point greater than
2193 127 whose first byte was different to the first byte of the other case of
2194 the letter, the other case of this starting letter was not recognized
2195 (#976).
2196
2197 12. If a pattern that was studied started with a repeated Unicode property
2198 test, for example, \p{Nd}+, there was the theoretical possibility of
2199 setting up an incorrect bitmap of starting bytes, but fortunately it could
2200 not have actually happened in practice until change 8 above was made (it
2201 added property types that matched character-matching opcodes).
2202
2203 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
2204 possible starting bytes for non-anchored patterns.
2205
2206 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
2207 \R, and also a number of cases that involve Unicode properties, both
2208 explicit and implicit when PCRE_UCP is set.
2209
2210 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
2211 input, it could crash or give wrong results if characters with values
2212 greater than 0xc0 were present in the subject string. (Detail: it assumed
2213 UTF-8 input when processing these items.)
2214
2215 16. Added a lot of (int) casts to avoid compiler warnings in systems where
2216 size_t is 64-bit (#991).
2217
2218 17. Added a check for running out of memory when PCRE is compiled with
2219 --disable-stack-for-recursion (#990).
2220
2221 18. If the last data line in a file for pcretest does not have a newline on
2222 the end, a newline was missing in the output.
2223
2224 19. The default pcre_chartables.c file recognizes only ASCII characters (values
2225 less than 128) in its various bitmaps. However, there is a facility for
2226 generating tables according to the current locale when PCRE is compiled. It
2227 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
2228 characters, are recognized by isspace() and therefore were getting set in
2229 these tables, and indeed these tables seem to approximate to ISO 8859. This
2230 caused a problem in UTF-8 mode when pcre_study() was used to create a list
2231 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
2232 which of course cannot start UTF-8 characters. I have changed the code so
2233 that only real ASCII characters (less than 128) and the correct starting
2234 bytes for UTF-8 encodings are set for characters greater than 127 when in
2235 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
2236 altogether.)
2237
2238 20. Added the /T option to pcretest so as to be able to run tests with non-
2239 standard character tables, thus making it possible to include the tests
2240 used for 19 above in the standard set of tests.
2241
2242 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
2243 reference to a subpattern the other side of a comment that contains an
2244 opening parenthesis caused either an internal compiling error, or a
2245 reference to the wrong subpattern.
2246
2247
2248 Version 8.02 19-Mar-2010
2249 ------------------------
2250
2251 1. The Unicode data tables have been updated to Unicode 5.2.0.
2252
2253 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
2254 configured.
2255
2256 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
2257 original author of that file, following a query about its status.
2258
2259 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
2260 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
2261
2262 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
2263 quantifier applied to a forward-referencing subroutine call, could compile
2264 incorrect code or give the error "internal error: previously-checked
2265 referenced subpattern not found".
2266
2267 6. Both MS Visual Studio and Symbian OS have problems with initializing
2268 variables to point to external functions. For these systems, therefore,
2269 pcre_malloc etc. are now initialized to local functions that call the
2270 relevant global functions.
2271
2272 7. There were two entries missing in the vectors called coptable and poptable
2273 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
2274 I've fixed the data, and added a kludgy way of testing at compile time that
2275 the lengths are correct (equal to the number of opcodes).
2276
2277 8. Following on from 7, I added a similar kludge to check the length of the
2278 eint vector in pcreposix.c.
2279
2280 9. Error texts for pcre_compile() are held as one long string to avoid too
2281 much relocation at load time. To find a text, the string is searched,
2282 counting zeros. There was no check for running off the end of the string,
2283 which could happen if a new error number was added without updating the
2284 string.
2285
2286 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
2287
2288 11. \K was not working if it appeared in an atomic group or in a group that
2289 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
2290 \K is "not well defined" if used in an assertion. PCRE now accepts it if
2291 the assertion is positive, but not if it is negative.
2292
2293 12. Change 11 fortuitously reduced the size of the stack frame used in the
2294 "match()" function of pcre_exec.c by one pointer. Forthcoming
2295 implementation of support for (*MARK) will need an extra pointer on the
2296 stack; I have reserved it now, so that the stack frame size does not
2297 decrease.
2298
2299 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
2300 item in branch that calls a recursion is a subroutine call - as in the
2301 second branch in the above example - was incorrectly given the compile-
2302 time error "recursive call could loop indefinitely" because pcre_compile()
2303 was not correctly checking the subroutine for matching a non-empty string.
2304
2305 14. The checks for overrunning compiling workspace could trigger after an
2306 overrun had occurred. This is a "should never occur" error, but it can be
2307 triggered by pathological patterns such as hundreds of nested parentheses.
2308 The checks now trigger 100 bytes before the end of the workspace.
2309
2310 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
2311
2312
2313 Version 8.01 19-Jan-2010
2314 ------------------------
2315
2316 1. If a pattern contained a conditional subpattern with only one branch (in
2317 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
2318 computed the wrong minimum data length (which is of course zero for such
2319 subpatterns). This could cause incorrect "no match" results.
2320
2321 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
2322 the pattern is reset in the first branch, pcre_compile() failed with
2323 "internal error: code overflow at offset...". This happened only when
2324 the reset was to the original external option setting. (An optimization
2325 abstracts leading options settings into an external setting, which was the
2326 cause of this.)
2327
2328 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
2329 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
2330 assertion pattern did not match (meaning that the assertion was true), it
2331 was incorrectly treated as false if the SKIP had been reached during the
2332 matching. This also applied to assertions used as conditions.
2333
2334 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
2335 assertion subpattern, including such a pattern used as a condition,
2336 unpredictable results occurred, instead of the error return
2337 PCRE_ERROR_DFA_UITEM.
2338
2339 5. The C++ GlobalReplace function was not working like Perl for the special
2340 situation when an empty string is matched. It now does the fancy magic
2341 stuff that is necessary.
2342
2343 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
2344 removed. (These were left over from very, very early versions of PCRE.)
2345
2346 7. Some cosmetic changes to the code to make life easier when compiling it
2347 as part of something else:
2348
2349 (a) Change DEBUG to PCRE_DEBUG.
2350
2351 (b) In pcre_compile(), rename the member of the "branch_chain" structure
2352 called "current" as "current_branch", to prevent a collision with the
2353 Linux macro when compiled as a kernel module.
2354
2355 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
2356 prevent a collision with the Linux macro when compiled as a kernel
2357 module.
2358
2359 8. In pcre_compile() there are some checks for integer overflows that used to
2360 cast potentially large values to (double). This has been changed to that
2361 when building, a check for int64_t is made, and if it is found, it is used
2362 instead, thus avoiding the use of floating point arithmetic. (There is no
2363 other use of FP in PCRE.) If int64_t is not found, the fallback is to
2364 double.
2365
2366 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
2367 2005 (difference between two addresses compared to an unsigned value).
2368
2369 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
2370 custom one, because of the following reported problem in Windows:
2371
2372 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
2373 under Win32.
2374 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
2375 therefore missing the function definition.
2376 - The compiler thus generates a "C" signature for the test function.
2377 - The linker fails to find the "C" function.
2378 - PCRE fails to configure if asked to do so against libbz2.
2379
2380 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
2381 messages were output:
2382
2383 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
2384 rerunning libtoolize, to keep the correct libtool macros in-tree.
2385 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
2386
2387 I have done both of these things.
2388
2389 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
2390 most of the time, it *can* run out if it is given a pattern that contains a
2391 runaway infinite recursion. I updated the discussion in the pcrestack man
2392 page.
2393
2394 13. Now that we have gone to the x.xx style of version numbers, the minor
2395 version may start with zero. Using 08 or 09 is a bad idea because users
2396 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
2397 interpreted as invalid octal numbers. I've updated the previous comment in
2398 configure.ac, and also added a check that gives an error if 08 or 09 are
2399 used.
2400
2401 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
2402 causing partial matching to fail when the end of the subject matched \W
2403 in a UTF-8 pattern where \W was quantified with a minimum of 3.
2404
2405 15. There were some discrepancies between the declarations in pcre_internal.h
2406 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
2407 their definitions. The declarations used "const uschar *" and the
2408 definitions used USPTR. Even though USPTR is normally defined as "const
2409 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
2410 reported that: "This difference in casting confuses some C++ compilers, for
2411 example, SunCC recognizes above declarations as different functions and
2412 generates broken code for hbpcre." I have changed the declarations to use
2413 USPTR.
2414
2415 16. GNU libtool is named differently on some systems. The autogen.sh script now
2416 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
2417 (FreeBSD).
2418
2419 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
2420 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
2421 comment: "Figure out how to create a longlong from a string: strtoll and
2422 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
2423 instance, but it only takes 2 args instead of 3!"
2424
2425 18. A subtle bug concerned with back references has been fixed by a change of
2426 specification, with a corresponding code fix. A pattern such as
2427 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
2428 refers, was giving matches when it shouldn't. For example, xa=xaaa would
2429 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
2430 same bug. Such groups have to be quantified to be useful, or contained
2431 inside another quantified group. (If there's no repetition, the reference
2432 can never match.) The problem arises because, having left the group and
2433 moved on to the rest of the pattern, a later failure that backtracks into
2434 the group uses the captured value from the final iteration of the group
2435 rather than the correct earlier one. I have fixed this in PCRE by forcing
2436 any group that contains a reference to itself to be an atomic group; that
2437 is, there cannot be any backtracking into it once it has completed. This is
2438 similar to recursive and subroutine calls.
2439
2440
2441 Version 8.00 19-Oct-09
2442 ----------------------
2443
2444 1. The table for translating pcre_compile() error codes into POSIX error codes
2445 was out-of-date, and there was no check on the pcre_compile() error code
2446 being within the table. This could lead to an OK return being given in
2447 error.
2448
2449 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
2450 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
2451 in a Windows environment.
2452
2453 3. The pcregrep --count option prints the count for each file even when it is
2454 zero, as does GNU grep. However, pcregrep was also printing all files when
2455 --files-with-matches was added. Now, when both options are given, it prints
2456 counts only for those files that have at least one match. (GNU grep just
2457 prints the file name in this circumstance, but including the count seems
2458 more useful - otherwise, why use --count?) Also ensured that the
2459 combination -clh just lists non-zero counts, with no names.
2460
2461 4. The long form of the pcregrep -F option was incorrectly implemented as
2462 --fixed_strings instead of --fixed-strings. This is an incompatible change,
2463 but it seems right to fix it, and I didn't think it was worth preserving
2464 the old behaviour.
2465
2466 5. The command line items --regex=pattern and --regexp=pattern were not
2467 recognized by pcregrep, which required --regex pattern or --regexp pattern
2468 (with a space rather than an '='). The man page documented the '=' forms,
2469 which are compatible with GNU grep; these now work.
2470
2471 6. No libpcreposix.pc file was created for pkg-config; there was just
2472 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
2473
2474 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
2475 when UCP support is not needed, by modifying the Python script that
2476 generates it from Unicode data files. This should not matter if the module
2477 is correctly used as a library, but I received one complaint about 50K of
2478 unwanted data. My guess is that the person linked everything into his
2479 program rather than using a library. Anyway, it does no harm.
2480
2481 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
2482 was a minimum greater than 1 for a wide character in a possessive
2483 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
2484 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
2485 character. Chaos in the form of incorrect output or a compiling loop could
2486 result.
2487
2488 9. The restrictions on what a pattern can contain when partial matching is
2489 requested for pcre_exec() have been removed. All patterns can now be
2490 partially matched by this function. In addition, if there are at least two
2491 slots in the offset vector, the offset of the earliest inspected character
2492 for the match and the offset of the end of the subject are set in them when
2493 PCRE_ERROR_PARTIAL is returned.
2494
2495 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
2496 synonymous with PCRE_PARTIAL, for backwards compatibility, and
2497 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
2498 and may be more useful for multi-segment matching.
2499
2500 11. Partial matching with pcre_exec() is now more intuitive. A partial match
2501 used to be given if ever the end of the subject was reached; now it is
2502 given only if matching could not proceed because another character was
2503 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
2504 string "Z", which now yields "no match" instead of "partial match". In the
2505 case of pcre_dfa_exec(), "no match" is given if every matching path for the
2506 final character ended with (*FAIL).
2507
2508 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
2509 if the pattern had a "must contain" character that was already found in the
2510 earlier partial match, unless partial matching was again requested. For
2511 example, with the pattern /dog.(body)?/, the "must contain" character is
2512 "g". If the first part-match was for the string "dog", restarting with
2513 "sbody" failed. This bug has been fixed.
2514
2515 13. The string returned by pcre_dfa_exec() after a partial match has been
2516 changed so that it starts at the first inspected character rather than the
2517 first character of the match. This makes a difference only if the pattern
2518 starts with a lookbehind assertion or \b or \B (\K is not supported by
2519 pcre_dfa_exec()). It's an incompatible change, but it makes the two
2520 matching functions compatible, and I think it's the right thing to do.
2521
2522 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
2523 so that the demonstration program is easily available in environments where
2524 PCRE has not been installed from source.
2525
2526 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
2527 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
2528 library.
2529
2530 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
2531 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
2532 is not the first non-POSIX option to be added. Clearly some people find
2533 these options useful.
2534
2535 17. If a caller to the POSIX matching function regexec() passes a non-zero
2536 value for nmatch with a NULL value for pmatch, the value of
2537 nmatch is forced to zero.
2538
2539 18. RunGrepTest did not have a test for the availability of the -u option of
2540 the diff command, as RunTest does. It now checks in the same way as
2541 RunTest, and also checks for the -b option.
2542
2543 19. If an odd number of negated classes containing just a single character
2544 interposed, within parentheses, between a forward reference to a named
2545 subpattern and the definition of the subpattern, compilation crashed with
2546 an internal error, complaining that it could not find the referenced
2547 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
2548 [The bug was that it was starting one character too far in when skipping
2549 over the character class, thus treating the ] as data rather than
2550 terminating the class. This meant it could skip too much.]
2551
2552 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
2553 /g option in pcretest when the pattern contains \K, which makes it possible
2554 to have an empty string match not at the start, even when the pattern is
2555 anchored. Updated pcretest and pcredemo to use this option.
2556
2557 21. If the maximum number of capturing subpatterns in a recursion was greater
2558 than the maximum at the outer level, the higher number was returned, but
2559 with unset values at the outer level. The correct (outer level) value is
2560 now given.
2561
2562 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
2563 PCRE did not set those parentheses (unlike Perl). I have now found a way to
2564 make it do so. The string so far is captured, making this feature
2565 compatible with Perl.
2566
2567 23. The tests have been re-organized, adding tests 11 and 12, to make it
2568 possible to check the Perl 5.10 features against Perl 5.10.
2569
2570 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
2571 pattern matches a fixed length string. PCRE did not allow this; now it
2572 does. Neither allows recursion.
2573
2574 25. I finally figured out how to implement a request to provide the minimum
2575 length of subject string that was needed in order to match a given pattern.
2576 (It was back references and recursion that I had previously got hung up
2577 on.) This code has now been added to pcre_study(); it finds a lower bound
2578 to the length of subject needed. It is not necessarily the greatest lower
2579 bound, but using it to avoid searching strings that are too short does give
2580 some useful speed-ups. The value is available to calling programs via
2581 pcre_fullinfo().
2582
2583 26. While implementing 25, I discovered to my embarrassment that pcretest had
2584 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
2585 study optimizations had never been tested with that matching function.
2586 Oops. What is worse, even when it was passed study data, there was a bug in
2587 pcre_dfa_exec() that meant it never actually used it. Double oops. There
2588 were also very few tests of studied patterns with pcre_dfa_exec().
2589
2590 27. If (?| is used to create subpatterns with duplicate numbers, they are now
2591 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
2592 on the other side of the coin, they are no longer allowed to have different
2593 names, because these cannot be distinguished in PCRE, and this has caused
2594 confusion. (This is a difference from Perl.)
2595
2596 28. When duplicate subpattern names are present (necessarily with different
2597 numbers, as required by 27 above), and a test is made by name in a
2598 conditional pattern, either for a subpattern having been matched, or for
2599 recursion in such a pattern, all the associated numbered subpatterns are
2600 tested, and the overall condition is true if the condition is true for any
2601 one of them. This is the way Perl works, and is also more like the way
2602 testing by number works.
2603
2604
2605 Version 7.9 11-Apr-09
2606 ---------------------
2607
2608 1. When building with support for bzlib/zlib (pcregrep) and/or readline
2609 (pcretest), all targets were linked against these libraries. This included
2610 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
2611 libraries. This caused unwanted dependencies to be created. This problem
2612 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
2613 pcretest is linked with readline.
2614
2615 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
2616 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
2617 moved outside it again, because FALSE and TRUE are already defined in AIX,
2618 but BOOL is not.
2619
2620 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
2621 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
2622
2623 4. The pcregrep documentation said spaces were inserted as well as colons (or
2624 hyphens) following file names and line numbers when outputting matching
2625 lines. This is not true; no spaces are inserted. I have also clarified the
2626 wording for the --colour (or --color) option.
2627
2628 5. In pcregrep, when --colour was used with -o, the list of matching strings
2629 was not coloured; this is different to GNU grep, so I have changed it to be
2630 the same.
2631
2632 6. When --colo(u)r was used in pcregrep, only the first matching substring in
2633 each matching line was coloured. Now it goes on to look for further matches
2634 of any of the test patterns, which is the same behaviour as GNU grep.
2635
2636 7. A pattern that could match an empty string could cause pcregrep to loop; it
2637 doesn't make sense to accept an empty string match in pcregrep, so I have
2638 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
2639 seems to be how GNU grep behaves. [But see later change 40 for release
2640 8.33.]
2641
2642 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
2643 start or after a newline", because the conditional assertion was not being
2644 correctly handled. The rule now is that both the assertion and what follows
2645 in the first alternative must satisfy the test.
2646
2647 9. If auto-callout was enabled in a pattern with a conditional group whose
2648 condition was an assertion, PCRE could crash during matching, both with
2649 pcre_exec() and pcre_dfa_exec().
2650
2651 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
2652 used for matching.
2653
2654 11. Unicode property support in character classes was not working for
2655 characters (bytes) greater than 127 when not in UTF-8 mode.
2656
2657 12. Added the -M command line option to pcretest.
2658
2659 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
2660
2661 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
2662
2663 16. Added comments and documentation about mis-use of no_arg in the C++
2664 wrapper.
2665
2666 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
2667 from Martin Jerabek that uses macro names for all relevant character and
2668 string constants.
2669
2670 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
2671 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
2672 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
2673 these, but not everybody uses configure.
2674
2675 19. A conditional group that had only one branch was not being correctly
2676 recognized as an item that could match an empty string. This meant that an
2677 enclosing group might also not be so recognized, causing infinite looping
2678 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
2679 with the subject "ab", where knowledge that the repeated group can match
2680 nothing is needed in order to break the loop.
2681
2682 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
2683 exec(), but without supplying a callout function, matching went wrong.
2684
2685 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
2686 leak if the size of the offset vector was greater than 30. When the vector
2687 is smaller, the saved offsets during recursion go onto a local stack
2688 vector, but for larger vectors malloc() is used. It was failing to free
2689 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
2690 error, in fact).
2691
2692 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
2693 heapframe that is used only when UTF-8 support is enabled. This caused no
2694 problem, but was untidy.
2695
2696 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
2697 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
2698 included within another project.
2699
2700 24. Steven Van Ingelgem's patches to add more options to the CMake support,
2701 slightly modified by me:
2702
2703 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
2704 not building pcregrep.
2705
2706 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
2707 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
2708
2709 25. Forward references, both numeric and by name, in patterns that made use of
2710 duplicate group numbers, could behave incorrectly or give incorrect errors,
2711 because when scanning forward to find the reference group, PCRE was not
2712 taking into account the duplicate group numbers. A pattern such as
2713 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
2714
2715 26. Changed a few more instances of "const unsigned char *" to USPTR, making
2716 the feature of a custom pointer more persuasive (as requested by a user).
2717
2718 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
2719 pcretest.c, inside #ifndefs, because it seems they are sometimes already
2720 pre-defined.
2721
2722 28. Added support for (*UTF8) at the start of a pattern.
2723
2724 29. Arrange for flags added by the "release type" setting in CMake to be shown
2725 in the configuration summary.
2726
2727
2728 Version 7.8 05-Sep-08
2729 ---------------------
2730
2731 1. Replaced UCP searching code with optimized version as implemented for Ad
2732 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
2733 stage table and inline lookup instead of a function, giving speed ups of 2
2734 to 5 times on some simple patterns that I tested. Permission was given to
2735 distribute the MultiStage2.py script that generates the tables (it's not in
2736 the tarball, but is in the Subversion repository).
2737
2738 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
2739 scripts.
2740
2741 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
2742 a group with a zero qualifier. The result of the study could be incorrect,
2743 or the function might crash, depending on the pattern.
2744
2745 4. Caseless matching was not working for non-ASCII characters in back
2746 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
2747 It now works when Unicode Property Support is available.
2748
2749 5. In pcretest, an escape such as \x{de} in the data was always generating
2750 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
2751 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
2752 truncation.
2753
2754 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
2755
2756 7. Added two (int) casts to pcregrep when printing the difference of two
2757 pointers, in case they are 64-bit values.
2758
2759 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
2760 test 2 if it fails.
2761
2762 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
2763 and a #define of that name to empty if it is not externally set. This is to
2764 allow users of MSVC to set it if necessary.
2765
2766 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
2767 the convenience functions in the pcre_get.c source file.
2768
2769 11. An option change at the start of a pattern that had top-level alternatives
2770 could cause overwriting and/or a crash. This command provoked a crash in
2771 some environments:
2772
2773 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
2774
2775 This potential security problem was recorded as CVE-2008-2371.
2776
2777 12. For a pattern where the match had to start at the beginning or immediately
2778 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
2779 pcre_dfa_exec() could read past the end of the passed subject if there was
2780 no match. To help with detecting such bugs (e.g. with valgrind), I modified
2781 pcretest so that it places the subject at the end of its malloc-ed buffer.
2782
2783 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
2784 exec() might read past the end of the data buffer in UTF-8 mode.
2785
2786 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
2787 the data contained the byte 0x85 as part of a UTF-8 character within its
2788 first line. This applied both to normal and DFA matching.
2789
2790 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
2791 /^[^d]*?$/8 failed to match "abc".
2792
2793 16. Added a missing copyright notice to pcrecpp_internal.h.
2794
2795 17. Make it more clear in the documentation that values returned from
2796 pcre_exec() in ovector are byte offsets, not character counts.
2797
2798 18. Tidied a few places to stop certain compilers from issuing warnings.
2799
2800 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
2801 supplied by Stefan Weber. I made a further small update for 7.8 because
2802 there is a change of source arrangements: the pcre_searchfuncs.c module is
2803 replaced by pcre_ucd.c.
2804
2805
2806 Version 7.7 07-May-08
2807 ---------------------
2808
2809 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
2810 a string to a long long, pretend we don't even have a long long." This is
2811 done by checking for the strtoq, strtoll, and _strtoi64 functions.
2812
2813 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
2814 pre-7.6 versions, which defined a global no_arg variable instead of putting
2815 it in the RE class. (See also #8 below.)
2816
2817 3. Remove a line of dead code, identified by coverity and reported by Nuno
2818 Lopes.
2819
2820 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
2821
2822 (1) The include/exclude patterns were being applied to the whole pathnames
2823 of files, instead of just to the final components.
2824
2825 (2) If there was more than one level of directory, the subdirectories were
2826 skipped unless they satisfied the include/exclude conditions. This is
2827 inconsistent with GNU grep (and could even be seen as contrary to the
2828 pcregrep specification - which I improved to make it absolutely clear).
2829 The action now is always to scan all levels of directory, and just
2830 apply the include/exclude patterns to regular files.
2831
2832 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
2833 --exclude_dir in the tests to avoid scanning .svn directories.
2834
2835 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
2836 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
2837 doesn't support NULs in patterns.
2838
2839 7. Added some missing "const"s to declarations of static tables in
2840 pcre_compile.c and pcre_dfa_exec.c.
2841
2842 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
2843 caused by fix #2 above. (Subsequently also a second patch to fix the
2844 first patch. And a third patch - this was a messy problem.)
2845
2846 9. Applied Craig's patch to remove the use of push_back().
2847
2848 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
2849 matching function regexec().
2850
2851 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
2852 which, however, unlike Perl's \g{...}, are subroutine calls, not back
2853 references. PCRE supports relative numbers with this syntax (I don't think
2854 Oniguruma does).
2855
2856 12. Previously, a group with a zero repeat such as (...){0} was completely
2857 omitted from the compiled regex. However, this means that if the group
2858 was called as a subroutine from elsewhere in the pattern, things went wrong
2859 (an internal error was given). Such groups are now left in the compiled
2860 pattern, with a new opcode that causes them to be skipped at execution
2861 time.
2862
2863 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
2864 to the way PCRE behaves:
2865
2866 (a) A lone ] character is dis-allowed (Perl treats it as data).
2867
2868 (b) A back reference to an unmatched subpattern matches an empty string
2869 (Perl fails the current match path).
2870
2871 (c) A data ] in a character class must be notated as \] because if the
2872 first data character in a class is ], it defines an empty class. (In
2873 Perl it is not possible to have an empty class.) The empty class []
2874 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
2875 The negative empty class [^] matches any one character, independently
2876 of the DOTALL setting.
2877
2878 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
2879 non-existent subpattern following a character class starting with ']' and
2880 containing () gave an internal compiling error instead of "reference to
2881 non-existent subpattern". Fortunately, when the pattern did exist, the
2882 compiled code was correct. (When scanning forwards to check for the
2883 existence of the subpattern, it was treating the data ']' as terminating
2884 the class, so got the count wrong. When actually compiling, the reference
2885 was subsequently set up correctly.)
2886
2887 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
2888 it was being rejected as not supported by pcre_dfa_exec(), even though
2889 other assertions are supported. I have made pcre_dfa_exec() support
2890 (*FAIL).
2891
2892 16. The implementation of 13c above involved the invention of a new opcode,
2893 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
2894 cannot be changed at match time, I realized I could make a small
2895 improvement to matching performance by compiling OP_ALLANY instead of
2896 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
2897 on the OP_ANY path.
2898
2899 17. Compiling pcretest on Windows with readline support failed without the
2900 following two fixes: (1) Make the unistd.h include conditional on
2901 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
2902
2903 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
2904 ncurses library to be included for pcretest when ReadLine support is
2905 requested, but also to allow for it to be overridden. This patch came from
2906 Daniel Bergström.
2907
2908 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
2909 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
2910 any errors with the current Unicode tables. Thanks to Peter Kankowski for
2911 spotting this.
2912
2913
2914 Version 7.6 28-Jan-08
2915 ---------------------
2916
2917 1. A character class containing a very large number of characters with
2918 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
2919 overflow.
2920
2921 2. Patch to cut out the "long long" test in pcrecpp_unittest when
2922 HAVE_LONG_LONG is not defined.
2923
2924 3. Applied Christian Ehrlicher's patch to update the CMake build files to
2925 bring them up to date and include new features. This patch includes:
2926
2927 - Fixed PH's badly added libz and libbz2 support.
2928 - Fixed a problem with static linking.
2929 - Added pcredemo. [But later removed - see 7 below.]
2930 - Fixed dftables problem and added an option.
2931 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
2932 HAVE_LONG_LONG.
2933 - Added readline support for pcretest.
2934 - Added an listing of the option settings after cmake has run.
2935
2936 4. A user submitted a patch to Makefile that makes it easy to create
2937 "pcre.dll" under mingw when using Configure/Make. I added stuff to
2938 Makefile.am that cause it to include this special target, without
2939 affecting anything else. Note that the same mingw target plus all
2940 the other distribution libraries and programs are now supported
2941 when configuring with CMake (see 6 below) instead of with
2942 Configure/Make.
2943
2944 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
2945 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
2946 exported in the Windows port". It has not yet been confirmed that the patch
2947 solves the problem, but it does no harm.
2948
2949 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
2950 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
2951 with CMake, and also correct the comment about stack recursion.
2952
2953 7. Remove the automatic building of pcredemo from the ./configure system and
2954 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
2955 of a program that users should build themselves after PCRE is installed, so
2956 building it automatically is not really right. What is more, it gave
2957 trouble in some build environments.
2958
2959 8. Further tidies to CMakeLists.txt from Sheri and Christian.
2960
2961
2962 Version 7.5 10-Jan-08
2963 ---------------------
2964
2965 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
2966 values in parens when parsing an RE using the C++ wrapper."
2967
2968 2. Negative specials like \S did not work in character classes in UTF-8 mode.
2969 Characters greater than 255 were excluded from the class instead of being
2970 included.
2971
2972 3. The same bug as (2) above applied to negated POSIX classes such as
2973 [:^space:].
2974
2975 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
2976 defined or documented. It seems to have been a typo for PCRE_STATIC, so
2977 I have changed it.
2978
2979 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
2980 first named subpattern) and a construct such as (?&a) would reference the
2981 first named subpattern whose name started with "a" (in other words, the
2982 length check was missing). Both these problems are fixed. "Subpattern name
2983 expected" is now given for (?&) (a zero-length name), and this patch also
2984 makes it give the same error for \k'' (previously it complained that that
2985 was a reference to a non-existent subpattern).
2986
2987 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
2988 this is right because (?- can be followed by option settings as well as by
2989 digits. I have, however, made the messages clearer.
2990
2991 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
2992 than the number used in the conditional) now cause a compile-time error.
2993 This is actually not compatible with Perl, which accepts such patterns, but
2994 treats the conditional as always being FALSE (as PCRE used to), but it
2995 seems to me that giving a diagnostic is better.
2996
2997 8. Change "alphameric" to the more common word "alphanumeric" in comments
2998 and messages.
2999
3000 9. Fix two occurrences of "backslash" in comments that should have been
3001 "backspace".
3002
3003 10. Remove two redundant lines of code that can never be obeyed (their function
3004 was moved elsewhere).
3005
3006 11. The program that makes PCRE's Unicode character property table had a bug
3007 which caused it to generate incorrect table entries for sequences of
3008 characters that have the same character type, but are in different scripts.
3009 It amalgamated them into a single range, with the script of the first of
3010 them. In other words, some characters were in the wrong script. There were
3011 thirteen such cases, affecting characters in the following ranges:
3012
3013 U+002b0 - U+002c1
3014 U+0060c - U+0060d
3015 U+0061e - U+00612
3016 U+0064b - U+0065e
3017 U+0074d - U+0076d
3018 U+01800 - U+01805
3019 U+01d00 - U+01d77
3020 U+01d9b - U+01dbf
3021 U+0200b - U+0200f
3022 U+030fc - U+030fe
3023 U+03260 - U+0327f
3024 U+0fb46 - U+0fbb1
3025 U+10450 - U+1049d
3026
3027 12. The -o option (show only the matching part of a line) for pcregrep was not
3028 compatible with GNU grep in that, if there was more than one match in a
3029 line, it showed only the first of them. It now behaves in the same way as
3030 GNU grep.
3031
3032 13. If the -o and -v options were combined for pcregrep, it printed a blank
3033 line for every non-matching line. GNU grep prints nothing, and pcregrep now
3034 does the same. The return code can be used to tell if there were any
3035 non-matching lines.
3036
3037 14. Added --file-offsets and --line-offsets to pcregrep.
3038
3039 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
3040 infinitely looping recursion. The bug was that positive lookaheads were not
3041 being skipped when checking for a possible empty match (negative lookaheads
3042 and both kinds of lookbehind were skipped).
3043
3044 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
3045 inclusion of <windows.h> to before rather than after the definition of
3046 INVALID_FILE_ATTRIBUTES (patch from David Byron).
3047
3048 17. Specifying a possessive quantifier with a specific limit for a Unicode
3049 character property caused pcre_compile() to compile bad code, which led at
3050 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
3051 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
3052 caused the error; without that there was no problem.
3053
3054 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
3055
3056 19. Added --enable-pcretest-libreadline.
3057
3058 20. In pcrecpp.cc, the variable 'count' was incremented twice in
3059 RE::GlobalReplace(). As a result, the number of replacements returned was
3060 double what it should be. I removed one of the increments, but Craig sent a
3061 later patch that removed the other one (the right fix) and added unit tests
3062 that check the return values (which was not done before).
3063
3064 21. Several CMake things:
3065
3066 (1) Arranged that, when cmake is used on Unix, the libraries end up with
3067 the names libpcre and libpcreposix, not just pcre and pcreposix.
3068
3069 (2) The above change means that pcretest and pcregrep are now correctly
3070 linked with the newly-built libraries, not previously installed ones.
3071
3072 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
3073
3074 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
3075 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
3076 UTF-8 newline character). The key issue is that the pattern starts .*;
3077 this means that the match must be either at the beginning, or after a
3078 newline. The bug was in the code for advancing after a failed match and
3079 checking that the new position followed a newline. It was not taking
3080 account of UTF-8 characters correctly.
3081
3082 23. PCRE was behaving differently from Perl in the way it recognized POSIX
3083 character classes. PCRE was not treating the sequence [:...:] as a
3084 character class unless the ... were all letters. Perl, however, seems to
3085 allow any characters between [: and :], though of course it rejects as
3086 unknown any "names" that contain non-letters, because all the known class
3087 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
3088 for example, whereas PCRE did not - it did not recognize a POSIX character
3089 class. This seemed a bit dangerous, so the code has been changed to be
3090 closer to Perl. The behaviour is not identical to Perl, because PCRE will
3091 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
3092 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
3093 Perl does, and where it didn't before.
3094
3095 24. Rewrite so as to remove the single use of %n from pcregrep because in some
3096 Windows environments %n is disabled by default.
3097
3098
3099 Version 7.4 21-Sep-07
3100 ---------------------
3101
3102 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
3103 means that a class such as [\s] counted as "explicit reference to CR or
3104 LF". That isn't really right - the whole point of the change was to try to
3105 help when there was an actual mention of one of the two characters. So now
3106 the change happens only if \r or \n (or a literal CR or LF) character is
3107 encountered.
3108
3109 2. The 32-bit options word was also used for 6 internal flags, but the numbers
3110 of both had grown to the point where there were only 3 bits left.
3111 Fortunately, there was spare space in the data structure, and so I have
3112 moved the internal flags into a new 16-bit field to free up more option
3113 bits.
3114
3115 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
3116 but did not set the internal JCHANGED flag - either of these is enough to
3117 control the way the "get" function works - but the PCRE_INFO_JCHANGED
3118 facility is supposed to tell if (?J) was ever used, so now (?J) at the
3119 start sets both bits.
3120
3121 4. Added options (at build time, compile time, exec time) to change \R from
3122 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
3123
3124 5. doc/pcresyntax.html was missing from the distribution.
3125
3126 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
3127 compatibility, even though it is no longer used.
3128
3129 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
3130 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
3131 windows.h file is present (where different names are used). [This was
3132 reversed later after testing - see 16 below.]
3133
3134 8. Changed all #include <config.h> to #include "config.h". There were also
3135 some further <pcre.h> cases that I changed to "pcre.h".
3136
3137 9. When pcregrep was used with the --colour option, it missed the line ending
3138 sequence off the lines that it output.
3139
3140 10. It was pointed out to me that arrays of string pointers cause lots of
3141 relocations when a shared library is dynamically loaded. A technique of
3142 using a single long string with a table of offsets can drastically reduce
3143 these. I have refactored PCRE in four places to do this. The result is
3144 dramatic:
3145
3146 Originally: 290
3147 After changing UCP table: 187
3148 After changing error message table: 43
3149 After changing table of "verbs" 36
3150 After changing table of Posix names 22
3151
3152 Thanks to the folks working on Gregex for glib for this insight.
3153
3154 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
3155 unicode-properties was also set.
3156
3157 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
3158
3159 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
3160 checked only for CRLF.
3161
3162 14. Added casts to pcretest.c to avoid compiler warnings.
3163
3164 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
3165
3166 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
3167 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
3168 entirely. This removes changes made in 7 above.
3169
3170 17. The CMake files have been updated, and there is now more information about
3171 building with CMake in the NON-UNIX-USE document.
3172
3173
3174 Version 7.3 28-Aug-07
3175 ---------------------
3176
3177 1. In the rejigging of the build system that eventually resulted in 7.1, the
3178 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
3179 brackets there is not right, since it causes compilers to look for an
3180 installed pcre.h, not the version that is in the source that is being
3181 compiled (which of course may be different). I have changed it back to:
3182
3183 #include "pcre.h"
3184
3185 I have a vague recollection that the change was concerned with compiling in
3186 different directories, but in the new build system, that is taken care of
3187 by the VPATH setting the Makefile.
3188
3189 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
3190 when the subject happened to end in the byte 0x85 (e.g. if the last
3191 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
3192 characters but of course it shouldn't be taken as a newline when it is part
3193 of another character. The bug was that, for an unlimited repeat of . in
3194 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
3195 characters when looking for a newline.
3196
3197 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
3198
3199 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
3200 in debug output.
3201
3202 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
3203 long printing in the pcrecpp unittest when running under MinGW.
3204
3205 6. ESC_K was left out of the EBCDIC table.
3206
3207 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
3208 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
3209 limit also applies to "virtual nesting" when a pattern is recursive, and in
3210 this case 1000 isn't so big. I have been able to remove this limit at the
3211 expense of backing off one optimization in certain circumstances. Normally,
3212 when pcre_exec() would call its internal match() function recursively and
3213 immediately return the result unconditionally, it uses a "tail recursion"
3214 feature to save stack. However, when a subpattern that can match an empty
3215 string has an unlimited repetition quantifier, it no longer makes this
3216 optimization. That gives it a stack frame in which to save the data for
3217 checking that an empty string has been matched. Previously this was taken
3218 from the 1000-entry workspace that had been reserved. So now there is no
3219 explicit limit, but more stack is used.
3220
3221 8. Applied Daniel's patches to solve problems with the import/export magic
3222 syntax that is required for Windows, and which was going wrong for the
3223 pcreposix and pcrecpp parts of the library. These were overlooked when this
3224 problem was solved for the main library.
3225
3226 9. There were some crude static tests to avoid integer overflow when computing
3227 the size of patterns that contain repeated groups with explicit upper
3228 limits. As the maximum quantifier is 65535, the maximum group length was
3229 set at 30,000 so that the product of these two numbers did not overflow a
3230 32-bit integer. However, it turns out that people want to use groups that
3231 are longer than 30,000 bytes (though not repeat them that many times).
3232 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
3233 made it possible to implement the integer overflow checks in a much more
3234 dynamic way, which I have now done. The artificial limitation on group
3235 length has been removed - we now have only the limit on the total length of
3236 the compiled pattern, which depends on the LINK_SIZE setting.
3237
3238 10. Fixed a bug in the documentation for get/copy named substring when
3239 duplicate names are permitted. If none of the named substrings are set, the
3240 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
3241 empty string.
3242
3243 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
3244 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
3245 because the ] is interpreted as the first data character and the
3246 terminating ] is not found. PCRE has been made compatible with Perl in this
3247 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
3248 cause memory overwriting.
3249
3250 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
3251 string has been matched (to stop an infinite loop). It was not recognizing
3252 a conditional subpattern that could match an empty string if that
3253 subpattern was within another subpattern. For example, it looped when
3254 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
3255 condition was not nested. This bug has been fixed.
3256
3257 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
3258 past the start of the subject in the presence of bytes with the top bit
3259 set, for example "\x8aBCD".
3260
3261 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
3262 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
3263
3264 14. Optimized (?!) to (*FAIL).
3265
3266 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
3267 This restricts code points to be within the range 0 to 0x10FFFF, excluding
3268 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
3269 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
3270 does: it's just the validity check that is more restrictive.
3271
3272 16. Inserted checks for integer overflows during escape sequence (backslash)
3273 processing, and also fixed erroneous offset values for syntax errors during
3274 backslash processing.
3275
3276 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
3277 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
3278
3279 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
3280 caused an overrun.
3281
3282 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
3283 something other than just ASCII characters) inside a group that had an
3284 unlimited repeat caused a loop at compile time (while checking to see
3285 whether the group could match an empty string).
3286
3287 20. Debugging a pattern containing \p or \P could cause a crash. For example,
3288 [\P{Any}] did so. (Error in the code for printing property names.)
3289
3290 21. An orphan \E inside a character class could cause a crash.
3291
3292 22. A repeated capturing bracket such as (A)? could cause a wild memory
3293 reference during compilation.
3294
3295 23. There are several functions in pcre_compile() that scan along a compiled
3296 expression for various reasons (e.g. to see if it's fixed length for look
3297 behind). There were bugs in these functions when a repeated \p or \P was
3298 present in the pattern. These operators have additional parameters compared
3299 with \d, etc, and these were not being taken into account when moving along
3300 the compiled data. Specifically:
3301
3302 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
3303 length.
3304
3305 (b) An item such as \pL+ within a repeated group could cause crashes or
3306 loops.
3307
3308 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
3309 "reference to non-existent subpattern" error.
3310
3311 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
3312
3313 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
3314 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
3315
3316 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
3317
3318 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
3319 character were causing crashes (broken optimization).
3320
3321 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
3322 \p or \P) caused a compile-time loop.
3323
3324 28. More problems have arisen in unanchored patterns when CRLF is a valid line
3325 break. For example, the unstudied pattern [\r\n]A does not match the string
3326 "\r\nA" because change 7.0/46 below moves the current point on by two
3327 characters after failing to match at the start. However, the pattern \nA
3328 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
3329 the same is true. There doesn't seem any very clean way out of this, but
3330 what I have chosen to do makes the common cases work: PCRE now takes note
3331 of whether there can be an explicit match for \r or \n anywhere in the
3332 pattern, and if so, 7.0/46 no longer applies. As part of this change,
3333 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
3334 pattern has explicit CR or LF references.
3335
3336 29. Added (*CR) etc for changing newline setting at start of pattern.
3337
3338
3339 Version 7.2 19-Jun-07
3340 ---------------------
3341
3342 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
3343 which is apparently normally available under Windows.
3344
3345 2. Re-jig the pcregrep tests with different newline settings in an attempt
3346 to make them independent of the local environment's newline setting.
3347
3348 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
3349
3350 4. Some of the "internals" tests were previously cut out when the link size
3351 was not 2, because the output contained actual offsets. The recent new
3352 "Z" feature of pcretest means that these can be cut out, making the tests
3353 usable with all link sizes.
3354
3355 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
3356 stack recursion. This gives a massive performance boost under BSD, but just
3357 a small improvement under Linux. However, it saves one field in the frame
3358 in all cases.
3359
3360 6. Added more features from the forthcoming Perl 5.10:
3361
3362 (a) (?-n) (where n is a string of digits) is a relative subroutine or
3363 recursion call. It refers to the nth most recently opened parentheses.
3364
3365 (b) (?+n) is also a relative subroutine call; it refers to the nth next
3366 to be opened parentheses.
3367
3368 (c) Conditions that refer to capturing parentheses can be specified
3369 relatively, for example, (?(-2)... or (?(+3)...
3370
3371 (d) \K resets the start of the current match so that everything before
3372 is not part of it.
3373
3374 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
3375
3376 (f) \g{name} is another synonym - part of Perl 5.10's unification of
3377 reference syntax.
3378
3379 (g) (?| introduces a group in which the numbering of parentheses in each
3380 alternative starts with the same number.
3381
3382 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
3383
3384 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
3385 PCRE_INFO_JCHANGED.
3386
3387 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
3388 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
3389 for detecting groups that can match an empty string.
3390
3391 9. A pattern with a very large number of alternatives (more than several
3392 hundred) was running out of internal workspace during the pre-compile
3393 phase, where pcre_compile() figures out how much memory will be needed. A
3394 bit of new cunning has reduced the workspace needed for groups with
3395 alternatives. The 1000-alternative test pattern now uses 12 bytes of
3396 workspace instead of running out of the 4096 that are available.
3397
3398 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
3399
3400 11. Applied patch from Google to remove an optimization that didn't quite work.
3401 The report of the bug said:
3402
3403 pcrecpp::RE("a*").FullMatch("aaa") matches, while
3404 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
3405 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
3406
3407 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
3408 it matched the wrong number of bytes.
3409
3410
3411 Version 7.1 24-Apr-07
3412 ---------------------
3413
3414 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
3415 that is more "standard", making use of automake and other Autotools. There
3416 is some re-arrangement of the files and adjustment of comments consequent
3417 on this.
3418
3419 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
3420 for recursive directory scanning broke on some systems because the files
3421 are not scanned in any specific order and on different systems the order
3422 was different. A call to "sort" has been inserted into RunGrepTest for the
3423 approprate test as a short-term fix. In the longer term there may be an
3424 alternative.
3425
3426 3. I had an email from Eric Raymond about problems translating some of PCRE's
3427 man pages to HTML (despite the fact that I distribute HTML pages, some
3428 people do their own conversions for various reasons). The problems
3429 concerned the use of low-level troff macros .br and .in. I have therefore
3430 removed all such uses from the man pages (some were redundant, some could
3431 be replaced by .nf/.fi pairs). The 132html script that I use to generate
3432 HTML has been updated to handle .nf/.fi and to complain if it encounters
3433 .br or .in.
3434
3435 4. Updated comments in configure.ac that get placed in config.h.in and also
3436 arranged for config.h to be included in the distribution, with the name
3437 config.h.generic, for the benefit of those who have to compile without
3438 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
3439
3440 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
3441 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
3442 makevp.bat for the current PCRE, using the additional files
3443 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
3444
3445 6. A Windows user reported a minor discrepancy with test 2, which turned out
3446 to be caused by a trailing space on an input line that had got lost in his
3447 copy. The trailing space was an accident, so I've just removed it.
3448
3449 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
3450 that is needed.
3451
3452 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
3453 as "const" (a) because they are and (b) because it helps the PHP
3454 maintainers who have recently made a script to detect big data structures
3455 in the php code that should be moved to the .rodata section. I remembered
3456 to update Builducptable as well, so it won't revert if ucptable.h is ever
3457 re-created.
3458
3459 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
3460 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
3461 order to be able to cut out the UTF-8 tables in the latter when UTF-8
3462 support is not required. This saves 1.5-2K of code, which is important in
3463 some applications.
3464
3465 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
3466 so as not to refer to the tables, even though these functions will never be
3467 called when UTF-8 support is disabled. Otherwise there are problems with a
3468 shared library.
3469
3470 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
3471
3472 (a) It was defining its arguments as char * instead of void *.
3473
3474 (b) It was assuming that all moves were upwards in memory; this was true
3475 a long time ago when I wrote it, but is no longer the case.
3476
3477 The emulated memove() is provided for those environments that have neither
3478 memmove() nor bcopy(). I didn't think anyone used it these days, but that
3479 is clearly not the case, as these two bugs were recently reported.
3480
3481 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
3482 and Detrail to create the HTML documentation, the .txt form of the man
3483 pages, and it removes trailing spaces from listed files. It also creates
3484 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
3485 case, it wraps all the #defines with #ifndefs. This script should be run
3486 before "make dist".
3487
3488 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
3489 with Unicode property support.
3490
3491 (a) For a maximizing quantifier, if the two different cases of the
3492 character were of different lengths in their UTF-8 codings (there are
3493 some cases like this - I found 11), and the matching function had to
3494 back up over a mixture of the two cases, it incorrectly assumed they
3495 were both the same length.
3496
3497 (b) When PCRE was configured to use the heap rather than the stack for
3498 recursion during matching, it was not correctly preserving the data for
3499 the other case of a UTF-8 character when checking ahead for a match
3500 while processing a minimizing repeat. If the check also involved
3501 matching a wide character, but failed, corruption could cause an
3502 erroneous result when trying to check for a repeat of the original
3503 character.
3504
3505 13. Some tidying changes to the testing mechanism:
3506
3507 (a) The RunTest script now detects the internal link size and whether there
3508 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
3509 values substituted by "configure". (The RunGrepTest script already did
3510 this for UTF-8.) The configure.ac script no longer substitutes the
3511 relevant variables.
3512
3513 (b) The debugging options /B and /D in pcretest show the compiled bytecode
3514 with length and offset values. This means that the output is different
3515 for different internal link sizes. Test 2 is skipped for link sizes
3516 other than 2 because of this, bypassing the problem. Unfortunately,
3517 there was also a test in test 3 (the locale tests) that used /B and
3518 failed for link sizes other than 2. Rather than cut the whole test out,
3519 I have added a new /Z option to pcretest that replaces the length and
3520 offset values with spaces. This is now used to make test 3 independent
3521 of link size. (Test 2 will be tidied up later.)
3522
3523 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
3524 segmentation fault instead of returning the appropriate error message.
3525
3526 15. In multiline mode when the newline sequence was set to "any", the pattern
3527 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
3528 This doesn't seem right; it now treats the CRLF combination as the line
3529 ending, and so does not match in that case. It's only a pattern such as ^$
3530 that would hit this one: something like ^ABC$ would have failed after \r
3531 and then tried again after \r\n.
3532
3533 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
3534 in an attempt to make files that differ only in their line terminators
3535 compare equal. This works on Linux.
3536
3537 17. Under certain error circumstances pcregrep might try to free random memory
3538 as it exited. This is now fixed, thanks to valgrind.
3539
3540 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
3541 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
3542 was because its rules for how to advance for /g after matching an empty
3543 string at the end of a line did not allow for this case. They now check for
3544 it specially.
3545
3546 20. pcretest is supposed to handle patterns and data of any length, by
3547 extending its buffers when necessary. It was getting this wrong when the
3548 buffer for a data line had to be extended.
3549
3550 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
3551 CRLF as a newline sequence.
3552
3553 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
3554 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
3555 I have nevertheless tidied it up.
3556
3557 23. Added some casts to kill warnings from HP-UX ia64 compiler.
3558
3559 24. Added a man page for pcre-config.
3560
3561
3562 Version 7.0 19-Dec-06
3563 ---------------------
3564
3565 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
3566 moving to gcc 4.1.1.
3567
3568 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
3569 sys/time.h, which is documented as needed for this function. It doesn't
3570 seem to matter on Linux, but it showed up on some releases of OS X.
3571
3572 3. It seems that there are systems where bytes whose values are greater than
3573 127 match isprint() in the "C" locale. The "C" locale should be the
3574 default when a C program starts up. In most systems, only ASCII printing
3575 characters match isprint(). This difference caused the output from pcretest
3576 to vary, making some of the tests fail. I have changed pcretest so that:
3577
3578 (a) When it is outputting text in the compiled version of a pattern, bytes
3579 other than 32-126 are always shown as hex escapes.
3580
3581 (b) When it is outputting text that is a matched part of a subject string,
3582 it does the same, unless a different locale has been set for the match
3583 (using the /L modifier). In this case, it uses isprint() to decide.
3584
3585 4. Fixed a major bug that caused incorrect computation of the amount of memory
3586 required for a compiled pattern when options that changed within the
3587 pattern affected the logic of the preliminary scan that determines the
3588 length. The relevant options are -x, and -i in UTF-8 mode. The result was
3589 that the computed length was too small. The symptoms of this bug were
3590 either the PCRE error "internal error: code overflow" from pcre_compile(),
3591 or a glibc crash with a message such as "pcretest: free(): invalid next
3592 size (fast)". Examples of patterns that provoked this bug (shown in
3593 pcretest format) are:
3594
3595 /(?-x: )/x
3596 /(?x)(?-x: \s*#\s*)/
3597 /((?i)[\x{c0}])/8
3598 /(?i:[\x{c0}])/8
3599
3600 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
3601 is now done differently.
3602
3603 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
3604 wrapper classes; (b) implement a new function in the C++ scanner that is
3605 more efficient than the old way of doing things because it avoids levels of
3606 recursion in the regex matching; (c) add a paragraph to the documentation
3607 for the FullMatch() function.
3608
3609 6. The escape sequence \n was being treated as whatever was defined as
3610 "newline". Not only was this contrary to the documentation, which states
3611 that \n is character 10 (hex 0A), but it also went horribly wrong when
3612 "newline" was defined as CRLF. This has been fixed.
3613
3614 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
3615 was being set to -1 for the "end of line" case (supposedly a value that no
3616 character can have). Though this value is never used (the check for end of
3617 line is "zero bytes in current character"), it caused compiler complaints.
3618 I've changed it to 0xffffffff.
3619
3620 8. In pcre_version.c, the version string was being built by a sequence of
3621 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
3622 string (as it is for production releases) called a macro with an empty
3623 argument. The C standard says the result of this is undefined. The gcc
3624 compiler treats it as an empty string (which was what was wanted) but it is
3625 reported that Visual C gives an error. The source has been hacked around to
3626 avoid this problem.
3627
3628 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
3629 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
3630 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
3631 of them did).
3632
3633 10. Originally, pcretest opened its input and output without "b"; then I was
3634 told that "b" was needed in some environments, so it was added for release
3635 5.0 to both the input and output. (It makes no difference on Unix-like
3636 systems.) Later I was told that it is wrong for the input on Windows. I've
3637 now abstracted the modes into two macros, to make it easier to fiddle with
3638 them, and removed "b" from the input mode under Windows.
3639
3640 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
3641
3642 12. Added -help and --help to pcretest as an official way of being reminded
3643 of the options.
3644
3645 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
3646 and pcrecpp.cc because they annoy compilers at high warning levels.
3647
3648 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
3649
3650 15. Fixed an occurrence of == in configure.ac that should have been = (shell
3651 scripts are not C programs :-) and which was not noticed because it works
3652 on Linux.
3653
3654 16. pcretest is supposed to handle any length of pattern and data line (as one
3655 line or as a continued sequence of lines) by extending its input buffer if
3656 necessary. This feature was broken for very long pattern lines, leading to
3657 a string of junk being passed to pcre_compile() if the pattern was longer
3658 than about 50K.
3659
3660 17. I have done a major re-factoring of the way pcre_compile() computes the
3661 amount of memory needed for a compiled pattern. Previously, there was code
3662 that made a preliminary scan of the pattern in order to do this. That was
3663 OK when PCRE was new, but as the facilities have expanded, it has become
3664 harder and harder to keep it in step with the real compile phase, and there
3665 have been a number of bugs (see for example, 4 above). I have now found a
3666 cunning way of running the real compile function in a "fake" mode that
3667 enables it to compute how much memory it would need, while actually only
3668 ever using a few hundred bytes of working memory and without too many
3669 tests of the mode. This should make future maintenance and development
3670 easier. A side effect of this work is that the limit of 200 on the nesting
3671 depth of parentheses has been removed (though this was never a serious
3672 limitation, I suspect). However, there is a downside: pcre_compile() now
3673 runs more slowly than before (30% or more, depending on the pattern). I
3674 hope this isn't a big issue. There is no effect on runtime performance.
3675
3676 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
3677 newline (only possible for the last line of a file) and it was a
3678 pattern that set a locale (followed by /Lsomething), pcretest crashed.
3679
3680 19. Added additional timing features to pcretest. (1) The -tm option now times
3681 matching only, not compiling. (2) Both -t and -tm can be followed, as a
3682 separate command line item, by a number that specifies the number of
3683 repeats to use when timing. The default is 50000; this gives better
3684 precision, but takes uncomfortably long for very large patterns.
3685
3686 20. Extended pcre_study() to be more clever in cases where a branch of a
3687 subpattern has no definite first character. For example, (a*|b*)[cd] would
3688 previously give no result from pcre_study(). Now it recognizes that the
3689 first character must be a, b, c, or d.
3690
3691 21. There was an incorrect error "recursive call could loop indefinitely" if
3692 a subpattern (or the entire pattern) that was being tested for matching an
3693 empty string contained only one non-empty item after a nested subpattern.
3694 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
3695 incorrectly, because the \d was being skipped in the check.
3696
3697 22. The pcretest program now has a new pattern option /B and a command line
3698 option -b, which is equivalent to adding /B to every pattern. This causes
3699 it to show the compiled bytecode, without the additional information that
3700 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
3701 is the same as /B/I).
3702
3703 23. A new optimization is now able automatically to treat some sequences such
3704 as a*b as a*+b. More specifically, if something simple (such as a character
3705 or a simple class like \d) has an unlimited quantifier, and is followed by
3706 something that cannot possibly match the quantified thing, the quantifier
3707 is automatically "possessified".
3708
3709 24. A recursive reference to a subpattern whose number was greater than 39
3710 went wrong under certain circumstances in UTF-8 mode. This bug could also
3711 have affected the operation of pcre_study().
3712
3713 25. Realized that a little bit of performance could be had by replacing
3714 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
3715
3716 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
3717
3718 27. Possessive quantifiers such as a++ were previously implemented by turning
3719 them into atomic groups such as ($>a+). Now they have their own opcodes,
3720 which improves performance. This includes the automatically created ones
3721 from 23 above.
3722
3723 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
3724 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
3725 the first matched character to be a colon. This applied both to named and
3726 numbered groups.
3727
3728 29. The ucpinternal.h header file was missing its idempotency #ifdef.
3729
3730 30. I was sent a "project" file called libpcre.a.dev which I understand makes
3731 building PCRE on Windows easier, so I have included it in the distribution.
3732
3733 31. There is now a check in pcretest against a ridiculously large number being
3734 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
3735 loop, the loop is abandoned.
3736
3737 32. Forward references to subpatterns in conditions such as (?(2)...) where
3738 subpattern 2 is defined later cause pcre_compile() to search forwards in
3739 the pattern for the relevant set of parentheses. This search went wrong
3740 when there were unescaped parentheses in a character class, parentheses
3741 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
3742
3743 33. "Subroutine" calls and backreferences were previously restricted to
3744 referencing subpatterns earlier in the regex. This restriction has now
3745 been removed.
3746
3747 34. Added a number of extra features that are going to be in Perl 5.10. On the
3748 whole, these are just syntactic alternatives for features that PCRE had
3749 previously implemented using the Python syntax or my own invention. The
3750 other formats are all retained for compatibility.
3751
3752 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
3753 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
3754 also .NET compatible.
3755
3756 (b) A recursion or subroutine call to a named group can now be defined as
3757 (?&name) as well as (?P>name).
3758
3759 (c) A backreference to a named group can now be defined as \k<name> or
3760 \k'name' as well as (?P=name). The new forms, as well as being in Perl
3761 5.10, are also .NET compatible.
3762
3763 (d) A conditional reference to a named group can now use the syntax
3764 (?(<name>) or (?('name') as well as (?(name).
3765
3766 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
3767 groups (named and numbered) that are never evaluated inline, but can be
3768 called as "subroutines" from elsewhere. In effect, the DEFINE condition
3769 is always false. There may be only one alternative in such a group.
3770
3771 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
3772 as the simple (?(R). The condition is true only if the most recent
3773 recursion is that of the given number or name. It does not search out
3774 through the entire recursion stack.
3775
3776 (g) The escape \gN or \g{N} has been added, where N is a positive or
3777 negative number, specifying an absolute or relative reference.
3778
3779 35. Tidied to get rid of some further signed/unsigned compiler warnings and
3780 some "unreachable code" warnings.
3781
3782 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
3783 things, this adds five new scripts.
3784
3785 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
3786 There were also incompatibilities regarding the handling of \Q..\E inside
3787 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
3788 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
3789
3790 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
3791 matches an empty string, and forcibly breaks the loop. There were bugs in
3792 this code in non-simple cases. For a pattern such as ^(a()*)* matched
3793 against aaaa the result was just "a" rather than "aaaa", for example. Two
3794 separate and independent bugs (that affected different cases) have been
3795 fixed.
3796
3797 39. Refactored the code to abolish the use of different opcodes for small
3798 capturing bracket numbers. This is a tidy that I avoided doing when I
3799 removed the limit on the number of capturing brackets for 3.5 back in 2001.
3800 The new approach is not only tidier, it makes it possible to reduce the
3801 memory needed to fix the previous bug (38).
3802
3803 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
3804 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
3805 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
3806 mode.
3807
3808 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
3809 report.
3810
3811 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
3812 copy construction and assignment in the C++ wrapper.
3813
3814 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
3815 couple of bugs that could have given wrong results in the "--newline=crlf"
3816 case.
3817
3818 44. Added a number of casts and did some reorganization of signed/unsigned int
3819 variables following suggestions from Dair Grant. Also renamed the variable
3820 "this" as "item" because it is a C++ keyword.
3821
3822 45. Arranged for dftables to add
3823
3824 #include "pcre_internal.h"
3825
3826 to pcre_chartables.c because without it, gcc 4.x may remove the array
3827 definition from the final binary if PCRE is built into a static library and
3828 dead code stripping is activated.
3829
3830 46. For an unanchored pattern, if a match attempt fails at the start of a
3831 newline sequence, and the newline setting is CRLF or ANY, and the next two
3832 characters are CRLF, advance by two characters instead of one.
3833
3834
3835 Version 6.7 04-Jul-06
3836 ---------------------
3837
3838 1. In order to handle tests when input lines are enormously long, pcretest has
3839 been re-factored so that it automatically extends its buffers when
3840 necessary. The code is crude, but this _is_ just a test program. The
3841 default size has been increased from 32K to 50K.
3842
3843 2. The code in pcre_study() was using the value of the re argument before
3844 testing it for NULL. (Of course, in any sensible call of the function, it
3845 won't be NULL.)
3846
3847 3. The memmove() emulation function in pcre_internal.h, which is used on
3848 systems that lack both memmove() and bcopy() - that is, hardly ever -
3849 was missing a "static" storage class specifier.
3850
3851 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
3852 containing an extended class (one that cannot be represented by a bitmap
3853 because it contains high-valued characters or Unicode property items, e.g.
3854 [\pZ]). Almost always one would set UTF-8 mode when processing such a
3855 pattern, but PCRE should not loop if you do not (it no longer does).
3856 [Detail: two cases were found: (a) a repeated subpattern containing an
3857 extended class; (b) a recursive reference to a subpattern that followed a
3858 previous extended class. It wasn't skipping over the extended class
3859 correctly when UTF-8 mode was not set.]
3860
3861 5. A negated single-character class was not being recognized as fixed-length
3862 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
3863 compile error "lookbehind assertion is not fixed length".
3864
3865 6. The RunPerlTest auxiliary script was showing an unexpected difference
3866 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
3867 write a Perl script that can interpret lines of an input file either as
3868 byte characters or as UTF-8, which is what "perltest" was being required to
3869 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
3870 can't do is switch easily at run time between having the "use utf8;" pragma
3871 or not. In the end, I fudged it by using the RunPerlTest script to insert
3872 "use utf8;" explicitly for the UTF-8 tests.
3873
3874 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
3875 the end of the subject string, contrary to the documentation and to what
3876 Perl does. This was true of both matching functions. Now it matches only at
3877 the start of the subject and immediately after *internal* newlines.
3878
3879 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
3880 a pointer to an int instead of a pointer to an unsigned long int. This
3881 caused problems on 64-bit systems.
3882
3883 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
3884 instance of the 'standard' template library not being so standard".
3885
3886 10. There was no check on the number of named subpatterns nor the maximum
3887 length of a subpattern name. The product of these values is used to compute
3888 the size of the memory block for a compiled pattern. By supplying a very
3889 long subpattern name and a large number of named subpatterns, the size
3890 computation could be caused to overflow. This is now prevented by limiting
3891 the length of names to 32 characters, and the number of named subpatterns
3892 to 10,000.
3893
3894 11. Subpatterns that are repeated with specific counts have to be replicated in
3895 the compiled pattern. The size of memory for this was computed from the
3896 length of the subpattern and the repeat count. The latter is limited to
3897 65535, but there was no limit on the former, meaning that integer overflow
3898 could in principle occur. The compiled length of a repeated subpattern is
3899 now limited to 30,000 bytes in order to prevent this.
3900
3901 12. Added the optional facility to have named substrings with the same name.
3902
3903 13. Added the ability to use a named substring as a condition, using the
3904 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
3905 are numbers (not recommended). Forward references are permitted.
3906
3907 14. Added forward references in named backreferences (if you see what I mean).
3908
3909 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
3910 pattern could run off the end of the subject. For example, the pattern
3911 "(?s)(.{1,5})"8 did this with the subject "ab".
3912
3913 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
3914 PCRE_CASELESS was set when matching characters that were quantified with ?
3915 or *.
3916
3917 17. A character class other than a single negated character that had a minimum
3918 but no maximum quantifier - for example [ab]{6,} - was not handled
3919 correctly by pce_dfa_exec(). It would match only one character.
3920
3921 18. A valid (though odd) pattern that looked like a POSIX character
3922 class but used an invalid character after [ (for example [[,abc,]]) caused
3923 pcre_compile() to give the error "Failed: internal error: code overflow" or
3924 in some cases to crash with a glibc free() error. This could even happen if
3925 the pattern terminated after [[ but there just happened to be a sequence of
3926 letters, a binary zero, and a closing ] in the memory that followed.
3927
3928 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
3929 over the years. Originally (before any Unicode support), just the bottom 8
3930 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
3931 output from "man perlunicode" includes this:
3932
3933 The regular expression compiler produces polymorphic opcodes. That
3934 is, the pattern adapts to the data and automatically switches to
3935 the Unicode character scheme when presented with Unicode data--or
3936 instead uses a traditional byte scheme when presented with byte
3937 data.
3938
3939 Sadly, a wide octal escape does not cause a switch, and in a string with
3940 no other multibyte characters, these octal escapes are treated as before.
3941 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
3942 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
3943 Unicode string.
3944
3945 I have not perpetrated such confusion in PCRE. Up till now, it took just
3946 the bottom 8 bits, as in old Perl. I have now made octal escapes with
3947 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
3948 translate to the appropriate multibyte character.
3949
3950 29. Applied some refactoring to reduce the number of warnings from Microsoft
3951 and Borland compilers. This has included removing the fudge introduced
3952 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
3953 a warning about an unused variable.
3954
3955 21. PCRE has not included VT (character 0x0b) in the set of whitespace
3956 characters since release 4.0, because Perl (from release 5.004) does not.
3957 [Or at least, is documented not to: some releases seem to be in conflict
3958 with the documentation.] However, when a pattern was studied with
3959 pcre_study() and all its branches started with \s, PCRE still included VT
3960 as a possible starting character. Of course, this did no harm; it just
3961 caused an unnecessary match attempt.
3962
3963 22. Removed a now-redundant internal flag bit that recorded the fact that case
3964 dependency changed within the pattern. This was once needed for "required
3965 byte" processing, but is no longer used. This recovers a now-scarce options
3966 bit. Also moved the least significant internal flag bit to the most-
3967 significant bit of the word, which was not previously used (hangover from
3968 the days when it was an int rather than a uint) to free up another bit for
3969 the future.
3970
3971 23. Added support for CRLF line endings as well as CR and LF. As well as the
3972 default being selectable at build time, it can now be changed at runtime
3973 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
3974 specify that it is scanning data with non-default line endings.
3975
3976 24. Changed the definition of CXXLINK to make it agree with the definition of
3977 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
3978
3979 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
3980 recursions. This makes a big different to stack usage for some patterns.
3981
3982 26. If a subpattern containing a named recursion or subroutine reference such
3983 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
3984 the space required for the compiled pattern went wrong and gave too small a
3985 value. Depending on the environment, this could lead to "Failed: internal
3986 error: code overflow at offset 49" or "glibc detected double free or
3987 corruption" errors.
3988
3989 27. Applied patches from Google (a) to support the new newline modes and (b) to
3990 advance over multibyte UTF-8 characters in GlobalReplace.
3991
3992 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
3993 difference for some implementation of PCRE in some Windows version.
3994
3995 29. Added some extra testing facilities to pcretest:
3996
3997 \q<number> in a data line sets the "match limit" value
3998 \Q<number> in a data line sets the "match recursion limt" value
3999 -S <number> sets the stack size, where <number> is in megabytes
4000
4001 The -S option isn't available for Windows.
4002
4003
4004 Version 6.6 06-Feb-06
4005 ---------------------
4006
4007 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
4008 in pcreposix.h. I have copied the definition from pcre.h.
4009
4010 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
4011 because pcre.h is no longer a built file.
4012
4013 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
4014 not normally included in the compiled code.
4015
4016
4017 Version 6.5 01-Feb-06
4018 ---------------------
4019
4020 1. When using the partial match feature with pcre_dfa_exec(), it was not
4021 anchoring the second and subsequent partial matches at the new starting
4022 point. This could lead to incorrect results. For example, with the pattern
4023 /1234/, partially matching against "123" and then "a4" gave a match.
4024
4025 2. Changes to pcregrep:
4026
4027 (a) All non-match returns from pcre_exec() were being treated as failures
4028 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
4029 error message is output. Some extra information is given for the
4030 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
4031 probably the only errors that are likely to be caused by users (by
4032 specifying a regex that has nested indefinite repeats, for instance).
4033 If there are more than 20 of these errors, pcregrep is abandoned.
4034
4035 (b) A binary zero was treated as data while matching, but terminated the
4036 output line if it was written out. This has been fixed: binary zeroes
4037 are now no different to any other data bytes.
4038
4039 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
4040 used to set a locale for matching. The --locale=xxxx long option has
4041 been added (no short equivalent) to specify a locale explicitly on the
4042 pcregrep command, overriding the environment variables.
4043
4044 (d) When -B was used with -n, some line numbers in the output were one less
4045 than they should have been.
4046
4047 (e) Added the -o (--only-matching) option.
4048
4049 (f) If -A or -C was used with -c (count only), some lines of context were
4050 accidentally printed for the final match.
4051
4052 (g) Added the -H (--with-filename) option.
4053
4054 (h) The combination of options -rh failed to suppress file names for files
4055 that were found from directory arguments.
4056
4057 (i) Added the -D (--devices) and -d (--directories) options.
4058
4059 (j) Added the -F (--fixed-strings) option.
4060
4061 (k) Allow "-" to be used as a file name for -f as well as for a data file.
4062
4063 (l) Added the --colo(u)r option.
4064
4065 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
4066 is not present by default.
4067
4068 3. A nasty bug was discovered in the handling of recursive patterns, that is,
4069 items such as (?R) or (?1), when the recursion could match a number of
4070 alternatives. If it matched one of the alternatives, but subsequently,
4071 outside the recursion, there was a failure, the code tried to back up into
4072 the recursion. However, because of the way PCRE is implemented, this is not
4073 possible, and the result was an incorrect result from the match.
4074
4075 In order to prevent this happening, the specification of recursion has
4076 been changed so that all such subpatterns are automatically treated as
4077 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
4078
4079 4. I had overlooked the fact that, in some locales, there are characters for
4080 which isalpha() is true but neither isupper() nor islower() are true. In
4081 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
4082 and ordfeminine) are like this. This affected the treatment of \w and \W
4083 when they appeared in character classes, but not when they appeared outside
4084 a character class. The bit map for "word" characters is now created
4085 separately from the results of isalnum() instead of just taking it from the
4086 upper, lower, and digit maps. (Plus the underscore character, of course.)
4087
4088 5. The above bug also affected the handling of POSIX character classes such as
4089 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
4090 permanent tables. Instead, the bit maps for such a class were previously
4091 created as the appropriate unions of the upper, lower, and digit bitmaps.
4092 Now they are created by subtraction from the [[:word:]] class, which has
4093 its own bitmap.
4094
4095 6. The [[:blank:]] character class matches horizontal, but not vertical space.
4096 It is created by subtracting the vertical space characters (\x09, \x0a,
4097 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
4098 subtraction was done in the overall bitmap for a character class, meaning
4099 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
4100 be recognized. This bug has been fixed.
4101
4102 7. Patches from the folks at Google:
4103
4104 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
4105 real life, but is still worth protecting against".
4106
4107 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
4108 regular expressions".
4109
4110 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
4111 have it.
4112
4113 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
4114 "configure" and the latter not, in order to fix a problem somebody had
4115 with compiling the Arg class on HP-UX.
4116
4117 (e) Improve the error-handling of the C++ wrapper a little bit.
4118
4119 (f) New tests for checking recursion limiting.
4120
4121 8. The pcre_memmove() function, which is used only if the environment does not
4122 have a standard memmove() function (and is therefore rarely compiled),
4123 contained two bugs: (a) use of int instead of size_t, and (b) it was not
4124 returning a result (though PCRE never actually uses the result).
4125
4126 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
4127 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
4128 returned instead of calling malloc() with an overflowing number that would
4129 most likely cause subsequent chaos.
4130
4131 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
4132
4133 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
4134 with this option is matched, the nmatch and pmatch options of regexec() are
4135 ignored.
4136
4137 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
4138 provided in case anyone wants to the the POSIX interface with UTF-8
4139 strings.
4140
4141 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
4142 C++ linking (needed for some HP-UX environments).
4143
4144 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
4145 (unused parameter) and in the pcre_printint() function (omitted "default"
4146 switch label when the default is to do nothing).
4147
4148 15. Added some code to make it possible, when PCRE is compiled as a C++
4149 library, to replace subject pointers for pcre_exec() with a smart pointer
4150 class, thus making it possible to process discontinuous strings.
4151
4152 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
4153 much the same function. They were added by different people who were trying
4154 to make PCRE easy to compile on non-Unix systems. It has been suggested
4155 that PCRE_EXPORT be abolished now that there is more automatic apparatus
4156 for compiling on Windows systems. I have therefore replaced it with
4157 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
4158 defaults to "extern" for C or "extern C" for C++, which works fine on
4159 Unix-like systems. It is now possible to override the value of PCRE_DATA_
4160 SCOPE with something explicit in config.h. In addition:
4161
4162 (a) pcreposix.h still had just "extern" instead of either of these macros;
4163 I have replaced it with PCRE_DATA_SCOPE.
4164
4165 (b) Functions such as _pcre_xclass(), which are internal to the library,
4166 but external in the C sense, all had PCRE_EXPORT in their definitions.
4167 This is apparently wrong for the Windows case, so I have removed it.
4168 (It makes no difference on Unix-like systems.)
4169
4170 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
4171 of recursive calls to match(). This is different to MATCH_LIMIT because
4172 that limits the total number of calls to match(), not all of which increase
4173 the depth of recursion. Limiting the recursion depth limits the amount of
4174 stack (or heap if NO_RECURSE is set) that is used. The default can be set
4175 when PCRE is compiled, and changed at run time. A patch from Google adds
4176 this functionality to the C++ interface.
4177
4178 18. Changes to the handling of Unicode character properties:
4179
4180 (a) Updated the table to Unicode 4.1.0.
4181
4182 (b) Recognize characters that are not in the table as "Cn" (undefined).
4183
4184 (c) I revised the way the table is implemented to a much improved format
4185 which includes recognition of ranges. It now supports the ranges that
4186 are defined in UnicodeData.txt, and it also amalgamates other
4187 characters into ranges. This has reduced the number of entries in the
4188 table from around 16,000 to around 3,000, thus reducing its size
4189 considerably. I realized I did not need to use a tree structure after
4190 all - a binary chop search is just as efficient. Having reduced the
4191 number of entries, I extended their size from 6 bytes to 8 bytes to
4192 allow for more data.
4193
4194 (d) Added support for Unicode script names via properties such as \p{Han}.
4195
4196 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
4197 matching that character.
4198
4199 20. When matching a repeated Unicode property with a minimum greater than zero,
4200 (for example \pL{2,}), PCRE could look past the end of the subject if it
4201 reached it while seeking the minimum number of characters. This could
4202 happen only if some of the characters were more than one byte long, because
4203 there is a check for at least the minimum number of bytes.
4204
4205 21. Refactored the implementation of \p and \P so as to be more general, to
4206 allow for more different types of property in future. This has changed the
4207 compiled form incompatibly. Anybody with saved compiled patterns that use
4208 \p or \P will have to recompile them.
4209
4210 22. Added "Any" and "L&" to the supported property types.
4211
4212 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
4213 but give a compile time error if the value is greater than 0xff.
4214
4215 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
4216 accidentally not being installed or uninstalled.
4217
4218 25. The pcre.h file was built from pcre.h.in, but the only changes that were
4219 made were to insert the current release number. This seemed silly, because
4220 it made things harder for people building PCRE on systems that don't run
4221 "configure". I have turned pcre.h into a distributed file, no longer built
4222 by "configure", with the version identification directly included. There is
4223 no longer a pcre.h.in file.
4224
4225 However, this change necessitated a change to the pcre-config script as
4226 well. It is built from pcre-config.in, and one of the substitutions was the
4227 release number. I have updated configure.ac so that ./configure now finds
4228 the release number by grepping pcre.h.
4229
4230 26. Added the ability to run the tests under valgrind.
4231
4232
4233 Version 6.4 05-Sep-05
4234 ---------------------
4235
4236 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
4237 "--" to be printed when multiple files were scanned, even when none of the
4238 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
4239 consider it to be a bug, and have restored the previous behaviour.
4240
4241 2. A couple of code tidies to get rid of compiler warnings.
4242
4243 3. The pcretest program used to cheat by referring to symbols in the library
4244 whose names begin with _pcre_. These are internal symbols that are not
4245 really supposed to be visible externally, and in some environments it is
4246 possible to suppress them. The cheating is now confined to including
4247 certain files from the library's source, which is a bit cleaner.
4248
4249 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
4250 file's purpose clearer.
4251
4252 5. Reorganized pcre_ucp_findchar().
4253
4254
4255 Version 6.3 15-Aug-05
4256 ---------------------
4257
4258 1. The file libpcre.pc.in did not have general read permission in the tarball.
4259
4260 2. There were some problems when building without C++ support:
4261
4262 (a) If C++ support was not built, "make install" and "make test" still
4263 tried to test it.
4264
4265 (b) There were problems when the value of CXX was explicitly set. Some
4266 changes have been made to try to fix these, and ...
4267
4268 (c) --disable-cpp can now be used to explicitly disable C++ support.
4269
4270 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
4271 backslash in a target when C++ was disabled. This confuses some
4272 versions of "make", apparently. Using an intermediate variable solves
4273 this. (Same for CPP_LOBJ.)
4274
4275 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
4276 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
4277 necessary on certain architectures.
4278
4279 4. Added a setting of -export-symbols-regex to the link command to remove
4280 those symbols that are exported in the C sense, but actually are local
4281 within the library, and not documented. Their names all begin with
4282 "_pcre_". This is not a perfect job, because (a) we have to except some
4283 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
4284 available (and never for static libraries). I have made a note to try to
4285 find a way round (a) in the future.
4286
4287
4288 Version 6.2 01-Aug-05
4289 ---------------------
4290
4291 1. There was no test for integer overflow of quantifier values. A construction
4292 such as {1111111111111111} would give undefined results. What is worse, if
4293 a minimum quantifier for a parenthesized subpattern overflowed and became
4294 negative, the calculation of the memory size went wrong. This could have
4295 led to memory overwriting.
4296
4297 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
4298
4299 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
4300 operating environments where this matters.
4301
4302 4. Applied Giuseppe Maxia's patch to add additional features for controlling
4303 PCRE options from within the C++ wrapper.
4304
4305 5. Named capturing subpatterns were not being correctly counted when a pattern
4306 was compiled. This caused two problems: (a) If there were more than 100
4307 such subpatterns, the calculation of the memory needed for the whole
4308 compiled pattern went wrong, leading to an overflow error. (b) Numerical
4309 back references of the form \12, where the number was greater than 9, were
4310 not recognized as back references, even though there were sufficient
4311 previous subpatterns.
4312
4313 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
4314 versions of gcc, e.g. 2.95.4.
4315
4316
4317 Version 6.1 21-Jun-05
4318 ---------------------
4319
4320 1. There was one reference to the variable "posix" in pcretest.c that was not
4321 surrounded by "#if !defined NOPOSIX".
4322
4323 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
4324 the cross-check on the old pcre_info() function, for the benefit of the
4325 cut-down version of PCRE that is currently imported into Exim.
4326
4327 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
4328 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
4329 patterns that start (?i) but I don't think that matters. The use of (?i) is
4330 just an example; this all applies to the other options as well.
4331
4332 4. Since libtool seems to echo the compile commands it is issuing, the output
4333 from "make" can be reduced a bit by putting "@" in front of each libtool
4334 compile command.
4335
4336 5. Patch from the folks at Google for configure.in to be a bit more thorough
4337 in checking for a suitable C++ installation before trying to compile the
4338 C++ stuff. This should fix a reported problem when a compiler was present,
4339 but no suitable headers.
4340
4341 6. The man pages all had just "PCRE" as their title. I have changed them to
4342 be the relevant file name. I have also arranged that these names are
4343 retained in the file doc/pcre.txt, which is a concatenation in text format
4344 of all the man pages except the little individual ones for each function.
4345
4346 7. The NON-UNIX-USE file had not been updated for the different set of source
4347 files that come with release 6. I also added a few comments about the C++
4348 wrapper.
4349
4350
4351 Version 6.0 07-Jun-05
4352 ---------------------
4353
4354 1. Some minor internal re-organization to help with my DFA experiments.
4355
4356 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
4357 didn't matter for the library itself when fully configured, but did matter
4358 when compiling without UCP support, or within Exim, where the ucp files are
4359 not imported.
4360
4361 3. Refactoring of the library code to split up the various functions into
4362 different source modules. The addition of the new DFA matching code (see
4363 below) to a single monolithic source would have made it really too
4364 unwieldy, quite apart from causing all the code to be include in a
4365 statically linked application, when only some functions are used. This is
4366 relevant even without the DFA addition now that patterns can be compiled in
4367 one application and matched in another.
4368
4369 The downside of splitting up is that there have to be some external
4370 functions and data tables that are used internally in different modules of
4371 the library but which are not part of the API. These have all had their
4372 names changed to start with "_pcre_" so that they are unlikely to clash
4373 with other external names.
4374
4375 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
4376 a different (DFA) algorithm. Although it is slower than the original
4377 function, it does have some advantages for certain types of matching
4378 problem.
4379
4380 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
4381 including restarting after a partial match.
4382
4383 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
4384 defined when compiling for Windows was sent to me. I have put it into the
4385 code, though I have no means of testing or verifying it.
4386
4387 7. Added the pcre_refcount() auxiliary function.
4388
4389 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
4390 match before or at the first newline in the subject string. In pcretest,
4391 the /f option on a pattern can be used to set this.
4392
4393 9. A repeated \w when used in UTF-8 mode with characters greater than 256
4394 would behave wrongly. This has been present in PCRE since release 4.0.
4395
4396 10. A number of changes to the pcregrep command:
4397
4398 (a) Refactored how -x works; insert ^(...)$ instead of setting
4399 PCRE_ANCHORED and checking the length, in preparation for adding
4400 something similar for -w.
4401
4402 (b) Added the -w (match as a word) option.
4403
4404 (c) Refactored the way lines are read and buffered so as to have more
4405 than one at a time available.
4406
4407 (d) Implemented a pcregrep test script.
4408
4409 (e) Added the -M (multiline match) option. This allows patterns to match
4410 over several lines of the subject. The buffering ensures that at least
4411 8K, or the rest of the document (whichever is the shorter) is available
4412 for matching (and similarly the previous 8K for lookbehind assertions).
4413
4414 (f) Changed the --help output so that it now says
4415
4416 -w, --word-regex(p)
4417
4418 instead of two lines, one with "regex" and the other with "regexp"
4419 because that confused at least one person since the short forms are the
4420 same. (This required a bit of code, as the output is generated
4421 automatically from a table. It wasn't just a text change.)
4422
4423 (g) -- can be used to terminate pcregrep options if the next thing isn't an
4424 option but starts with a hyphen. Could be a pattern or a path name
4425 starting with a hyphen, for instance.
4426
4427 (h) "-" can be given as a file name to represent stdin.
4428
4429 (i) When file names are being printed, "(standard input)" is used for
4430 the standard input, for compatibility with GNU grep. Previously
4431 "<stdin>" was used.
4432
4433 (j) The option --label=xxx can be used to supply a name to be used for
4434 stdin when file names are being printed. There is no short form.
4435
4436 (k) Re-factored the options decoding logic because we are going to add
4437 two more options that take data. Such options can now be given in four
4438 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
4439
4440 (l) Added the -A, -B, and -C options for requesting that lines of context
4441 around matches be printed.
4442
4443 (m) Added the -L option to print the names of files that do not contain
4444 any matching lines, that is, the complement of -l.
4445
4446 (n) The return code is 2 if any file cannot be opened, but pcregrep does
4447 continue to scan other files.
4448
4449 (o) The -s option was incorrectly implemented. For compatibility with other
4450 greps, it now suppresses the error message for a non-existent or non-
4451 accessible file (but not the return code). There is a new option called
4452 -q that suppresses the output of matching lines, which was what -s was
4453 previously doing.
4454
4455 (p) Added --include and --exclude options to specify files for inclusion
4456 and exclusion when recursing.
4457
4458 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
4459 Hopefully, it now does.
4460
4461 12. Missing cast in pcre_study().
4462
4463 13. Added an "uninstall" target to the makefile.
4464
4465 14. Replaced "extern" in the function prototypes in Makefile.in with
4466 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
4467 world, but is set differently for Windows.
4468
4469 15. Added a second compiling function called pcre_compile2(). The only
4470 difference is that it has an extra argument, which is a pointer to an
4471 integer error code. When there is a compile-time failure, this is set
4472 non-zero, in addition to the error test pointer being set to point to an
4473 error message. The new argument may be NULL if no error number is required
4474 (but then you may as well call pcre_compile(), which is now just a
4475 wrapper). This facility is provided because some applications need a
4476 numeric error indication, but it has also enabled me to tidy up the way
4477 compile-time errors are handled in the POSIX wrapper.
4478
4479 16. Added VPATH=.libs to the makefile; this should help when building with one
4480 prefix path and installing with another. (Or so I'm told by someone who
4481 knows more about this stuff than I do.)
4482
4483 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
4484 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
4485 match everything, including newlines. This is not POSIX-compatible, but
4486 somebody wanted the feature. From pcretest it can be activated by using
4487 both the P and the s flags.
4488
4489 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
4490
4491 19. libpcre.pc was being incorrectly installed as executable.
4492
4493 20. A couple of places in pcretest check for end-of-line by looking for '\n';
4494 it now also looks for '\r' so that it will work unmodified on Windows.
4495
4496 21. Added Google's contributed C++ wrapper to the distribution.
4497
4498 22. Added some untidy missing memory free() calls in pcretest, to keep
4499 Electric Fence happy when testing.
4500
4501
4502
4503 Version 5.0 13-Sep-04
4504 ---------------------
4505
4506 1. Internal change: literal characters are no longer packed up into items
4507 containing multiple characters in a single byte-string. Each character
4508 is now matched using a separate opcode. However, there may be more than one
4509 byte in the character in UTF-8 mode.
4510
4511 2. The pcre_callout_block structure has two new fields: pattern_position and
4512 next_item_length. These contain the offset in the pattern to the next match
4513 item, and its length, respectively.
4514
4515 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
4516 insertion of callouts before each pattern item. Added the /C option to
4517 pcretest to make use of this.
4518
4519 4. On the advice of a Windows user, the lines
4520
4521 #if defined(_WIN32) || defined(WIN32)
4522 _setmode( _fileno( stdout ), 0x8000 );
4523 #endif /* defined(_WIN32) || defined(WIN32) */
4524
4525 have been added to the source of pcretest. This apparently does useful
4526 magic in relation to line terminators.
4527
4528 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
4529 for the benefit of those environments where the "b" makes a difference.
4530
4531 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
4532 to know about it. I have put a hack into configure.in that adds in code
4533 to set GCC=yes if CC=icc. This seems to end up at a point in the
4534 generated configure script that is early enough to affect the setting of
4535 compiler options, which is what is needed, but I have no means of testing
4536 whether it really works. (The user who reported this had patched the
4537 generated configure script, which of course I cannot do.)
4538
4539 LATER: After change 22 below (new libtool files), the configure script
4540 seems to know about icc (and also ecc). Therefore, I have commented out
4541 this hack in configure.in.
4542
4543 7. Added support for pkg-config (2 patches were sent in).
4544
4545 8. Negated POSIX character classes that used a combination of internal tables
4546 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
4547 [[:^ascii]]. Typically, they would match almost any characters. The other
4548 POSIX classes were not broken in this way.
4549
4550 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
4551 to find the match, as PCRE was deluded into thinking that the match had to
4552 start at the start point or following a newline. The same bug applied to
4553 patterns with negative forward assertions or any backward assertions
4554 preceding ".*" at the start, unless the pattern required a fixed first
4555 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
4556
4557 10. In UTF-8 mode, when moving forwards in the subject after a failed match
4558 starting at the last subject character, bytes beyond the end of the subject
4559 string were read.
4560
4561 11. Renamed the variable "class" as "classbits" to make life easier for C++
4562 users. (Previously there was a macro definition, but it apparently wasn't
4563 enough.)
4564
4565 12. Added the new field "tables" to the extra data so that tables can be passed
4566 in at exec time, or the internal tables can be re-selected. This allows
4567 a compiled regex to be saved and re-used at a later time by a different
4568 program that might have everything at different addresses.
4569
4570 13. Modified the pcre-config script so that, when run on Solaris, it shows a
4571 -R library as well as a -L library.
4572
4573 14. The debugging options of pcretest (-d on the command line or D on a
4574 pattern) showed incorrect output for anything following an extended class
4575 that contained multibyte characters and which was followed by a quantifier.
4576
4577 15. Added optional support for general category Unicode character properties
4578 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
4579 support. It adds about 90K to the size of the library. The meanings of the
4580 inbuilt class escapes such as \d and \s have NOT been changed.
4581
4582 16. Updated pcredemo.c to include calls to free() to release the memory for the
4583 compiled pattern.
4584
4585 17. The generated file chartables.c was being created in the source directory
4586 instead of in the building directory. This caused the build to fail if the
4587 source directory was different from the building directory, and was
4588 read-only.
4589
4590 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
4591 file. No doubt somebody will tell me if they don't make sense... Also added
4592 Dan Mooney's comments about building on OpenVMS.
4593
4594 19. Added support for partial matching via the PCRE_PARTIAL option for
4595 pcre_exec() and the \P data escape in pcretest.
4596
4597 20. Extended pcretest with 3 new pattern features:
4598
4599 (i) A pattern option of the form ">rest-of-line" causes pcretest to
4600 write the compiled pattern to the file whose name is "rest-of-line".
4601 This is a straight binary dump of the data, with the saved pointer to
4602 the character tables forced to be NULL. The study data, if any, is
4603 written too. After writing, pcretest reads a new pattern.
4604
4605 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
4606 compiled pattern from the given file. There must not be any
4607 occurrences of "<" in the file name (pretty unlikely); if there are,
4608 pcretest will instead treat the initial "<" as a pattern delimiter.
4609 After reading in the pattern, pcretest goes on to read data lines as
4610 usual.
4611
4612 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
4613 and 16-bit fields in a compiled pattern, to simulate a pattern that
4614 was compiled on a host of opposite endianness.
4615
4616 21. The pcre-exec() function can now cope with patterns that were compiled on
4617 hosts of opposite endianness, with this restriction:
4618
4619 As for any compiled expression that is saved and used later, the tables
4620 pointer field cannot be preserved; the extra_data field in the arguments
4621 to pcre_exec() should be used to pass in a tables address if a value
4622 other than the default internal tables were used at compile time.
4623
4624 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
4625 now diagnosed as an error. Previously, most of the time, a negative number
4626 would have been treated as zero, but if in addition "ovector" was passed as
4627 NULL, a crash could occur.
4628
4629 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
4630 new versions from the libtool 1.5 distribution (the last one is a copy of
4631 a file called libtool.m4). This seems to have fixed the need to patch
4632 "configure" to support Darwin 1.3 (which I used to do). However, I still
4633 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
4634 workstation).
4635
4636 24. Changed the PCRE licence to be the more standard "BSD" licence.
4637
4638
4639 Version 4.5 01-Dec-03
4640 ---------------------
4641
4642 1. There has been some re-arrangement of the code for the match() function so
4643 that it can be compiled in a version that does not call itself recursively.
4644 Instead, it keeps those local variables that need separate instances for
4645 each "recursion" in a frame on the heap, and gets/frees frames whenever it
4646 needs to "recurse". Keeping track of where control must go is done by means
4647 of setjmp/longjmp. The whole thing is implemented by a set of macros that
4648 hide most of the details from the main code, and operates only if
4649 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
4650 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
4651 operating.
4652
4653 To make it easier for callers to provide specially tailored get/free
4654 functions for this usage, two new functions, pcre_stack_malloc, and
4655 pcre_stack_free, are used. They are always called in strict stacking order,
4656 and the size of block requested is always the same.
4657
4658 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
4659 PCRE has been compiled to use the stack or the heap for recursion. The
4660 -C option of pcretest uses this to show which version is compiled.
4661
4662 A new data escape \S, is added to pcretest; it causes the amounts of store
4663 obtained and freed by both kinds of malloc/free at match time to be added
4664 to the output.
4665
4666 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
4667 what's available on my current Linux desktop machine.
4668
4669 3. When matching a UTF-8 string, the test for a valid string at the start has
4670 been extended. If start_offset is not zero, PCRE now checks that it points
4671 to a byte that is the start of a UTF-8 character. If not, it returns
4672 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
4673 this is necessary because there may be backward assertions in the pattern.
4674 When matching the same subject several times, it may save resources to use
4675 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
4676
4677 4. The code for checking the validity of UTF-8 strings has been tightened so
4678 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
4679 containing "overlong sequences".
4680
4681 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
4682 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
4683 should have been "&", but it just so happened that all the cases this let
4684 through by mistake were picked up later in the function.
4685
4686 6. I had used a variable called "isblank" - this is a C99 function, causing
4687 some compilers to warn. To avoid this, I renamed it (as "blankclass").
4688
4689 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
4690 prompting; (b) run "./pcretest /dev/null" at the start of the test script
4691 so the version is shown; (c) stop "make test" echoing "./RunTest".
4692
4693 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
4694
4695 9. The prototype for memmove() for systems that don't have it was using
4696 size_t, but the inclusion of the header that defines size_t was later. I've
4697 moved the #includes for the C headers earlier to avoid this.
4698
4699 10. Added some adjustments to the code to make it easier to compiler on certain
4700 special systems:
4701
4702 (a) Some "const" qualifiers were missing.
4703 (b) Added the macro EXPORT before all exported functions; by default this
4704 is defined to be empty.
4705 (c) Changed the dftables auxiliary program (that builds chartables.c) so
4706 that it reads its output file name as an argument instead of writing
4707 to the standard output and assuming this can be redirected.
4708
4709 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
4710 class containing characters with values greater than 255, PCRE compilation
4711 went into a loop.
4712
4713 12. A recursive reference to a subpattern that was within another subpattern
4714 that had a minimum quantifier of zero caused PCRE to crash. For example,
4715 (x(y(?2))z)? provoked this bug with a subject that got as far as the
4716 recursion. If the recursively-called subpattern itself had a zero repeat,
4717 that was OK.
4718
4719 13. In pcretest, the buffer for reading a data line was set at 30K, but the
4720 buffer into which it was copied (for escape processing) was still set at
4721 1024, so long lines caused crashes.
4722
4723 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
4724 "internal error: code overflow...". This applied to any character class
4725 that was followed by a possessive quantifier.
4726
4727 15. Modified the Makefile to add libpcre.la as a prerequisite for
4728 libpcreposix.la because I was told this is needed for a parallel build to
4729 work.
4730
4731 16. If a pattern that contained .* following optional items at the start was
4732 studied, the wrong optimizing data was generated, leading to matching
4733 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
4734 matching string must start with a or b or c. The correct conclusion for
4735 this pattern is that a match can start with any character.
4736
4737
4738 Version 4.4 13-Aug-03
4739 ---------------------
4740
4741 1. In UTF-8 mode, a character class containing characters with values between
4742 127 and 255 was not handled correctly if the compiled pattern was studied.
4743 In fixing this, I have also improved the studying algorithm for such
4744 classes (slightly).
4745
4746 2. Three internal functions had redundant arguments passed to them. Removal
4747 might give a very teeny performance improvement.
4748
4749 3. Documentation bug: the value of the capture_top field in a callout is *one
4750 more than* the number of the hightest numbered captured substring.
4751
4752 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
4753 in incorrectly linking with a previously installed version. They now link
4754 explicitly with libpcre.la.
4755
4756 5. configure.in no longer needs to recognize Cygwin specially.
4757
4758 6. A problem in pcre.in for Windows platforms is fixed.
4759
4760 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
4761 pcretest, it used to include the size of the study block as part of its
4762 output. Unfortunately, the structure contains a field that has a different
4763 size on different hardware architectures. This meant that the tests that
4764 showed this size failed. As the block is currently always of a fixed size,
4765 this information isn't actually particularly useful in pcretest output, so
4766 I have just removed it.
4767
4768 8. Three pre-processor statements accidentally did not start in column 1.
4769 Sadly, there are *still* compilers around that complain, even though
4770 standard C has not required this for well over a decade. Sigh.
4771
4772 9. In pcretest, the code for checking callouts passed small integers in the
4773 callout_data field, which is a void * field. However, some picky compilers
4774 complained about the casts involved for this on 64-bit systems. Now
4775 pcretest passes the address of the small integer instead, which should get
4776 rid of the warnings.
4777
4778 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
4779 both compile and run time, and gives an error if an invalid UTF-8 sequence
4780 is found. There is a option for disabling this check in cases where the
4781 string is known to be correct and/or the maximum performance is wanted.
4782
4783 11. In response to a bug report, I changed one line in Makefile.in from
4784
4785 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
4786 to
4787 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
4788
4789 to look similar to other lines, but I have no way of telling whether this
4790 is the right thing to do, as I do not use Windows. No doubt I'll get told
4791 if it's wrong...
4792
4793
4794 Version 4.3 21-May-03
4795 ---------------------
4796
4797 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
4798 Makefile.
4799
4800 2. Some refactoring to improve the quality of the code:
4801
4802 (i) The utf8_table... variables are now declared "const".
4803
4804 (ii) The code for \cx, which used the "case flipping" table to upper case
4805 lower case letters, now just substracts 32. This is ASCII-specific,
4806 but the whole concept of \cx is ASCII-specific, so it seems
4807 reasonable.
4808
4809 (iii) PCRE was using its character types table to recognize decimal and
4810 hexadecimal digits in the pattern. This is silly, because it handles
4811 only 0-9, a-f, and A-F, but the character types table is locale-
4812 specific, which means strange things might happen. A private
4813 table is now used for this - though it costs 256 bytes, a table is
4814 much faster than multiple explicit tests. Of course, the standard
4815 character types table is still used for matching digits in subject
4816 strings against \d.
4817
4818 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
4819 ending in _t are). So I've renamed it as ESC_tee.
4820
4821 3. The first argument for regexec() in the POSIX wrapper should have been
4822 defined as "const".
4823
4824 4. Changed pcretest to use malloc() for its buffers so that they can be
4825 Electric Fenced for debugging.
4826
4827 5. There were several places in the code where, in UTF-8 mode, PCRE would try
4828 to read one or more bytes before the start of the subject string. Often this
4829 had no effect on PCRE's behaviour, but in some circumstances it could
4830 provoke a segmentation fault.
4831
4832 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
4833 to try to read one or more bytes before the start of the subject string.
4834
4835 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
4836 UTF-8 support could misbehave in various ways if the subject string
4837 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
4838 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
4839 back over UTF-8 characters.)
4840
4841
4842 Version 4.2 14-Apr-03
4843 ---------------------
4844
4845 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
4846
4847 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
4848 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
4849 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
4850 [WIN_PREFIX]: new variable, "cyg" for Cygwin
4851 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
4852 and BUILD_EXEEXT
4853 Note: automatic setting of the BUILD variables is not yet working
4854 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
4855 compile-time but not at link-time
4856 [LINK]: use for linking executables only
4857 make different versions for Windows and non-Windows
4858 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
4859 libraries
4860 [LINK_FOR_BUILD]: new variable
4861 [OBJEXT]: use throughout
4862 [EXEEXT]: use throughout
4863 <winshared>: new target
4864 <wininstall>: new target
4865 <dftables.o>: use native compiler
4866 <dftables>: use native linker
4867 <install>: handle Windows platform correctly
4868 <clean>: ditto
4869 <check>: ditto
4870 copy DLL to top builddir before testing
4871
4872 As part of these changes, -no-undefined was removed again. This was reported
4873 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
4874 in any case.
4875
4876 3. Some tidies to get rid of compiler warnings:
4877
4878 . In the match_data structure, match_limit was an unsigned long int, whereas
4879 match_call_count was an int. I've made them both unsigned long ints.
4880
4881 . In pcretest the fact that a const uschar * doesn't automatically cast to
4882 a void * provoked a warning.
4883
4884 . Turning on some more compiler warnings threw up some "shadow" variables
4885 and a few more missing casts.
4886
4887 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
4888 option, a class that contained a single character with a value between 128
4889 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
4890
4891 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
4892 option, a class that contained several characters, but with at least one
4893 whose value was between 128 and 255 caused PCRE to crash.
4894
4895
4896 Version 4.1 12-Mar-03
4897 ---------------------
4898
4899 1. Compiling with gcc -pedantic found a couple of places where casts were
4900 needed, and a string in dftables.c that was longer than standard compilers are
4901 required to support.
4902
4903 2. Compiling with Sun's compiler found a few more places where the code could
4904 be tidied up in order to avoid warnings.
4905
4906 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
4907 first of these names is deprecated in the latest Autoconf in favour of the name
4908 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
4909 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
4910 analogy I have changed it to CFLAGS_FOR_BUILD.
4911
4912 4. Added -no-undefined to the linking command in the Makefile, because this is
4913 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
4914 linking step for the pcreposix library.
4915
4916 5. PCRE was failing to diagnose the case of two named groups with the same
4917 name.
4918
4919 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
4920 literal character that is needed in the subject for a match, and scans along to
4921 ensure that it is present before embarking on the full matching process. This
4922 saves time in cases of nested unlimited repeats that are never going to match.
4923 Problem: the scan can take a lot of time if the subject is very long (e.g.
4924 megabytes), thus penalizing straightforward matches. It is now done only if the
4925 amount of subject to be scanned is less than 1000 bytes.
4926
4927 7. A lesser problem with the same optimization is that it was recording the
4928 first character of an anchored pattern as "needed", thus provoking a search
4929 right along the subject, even when the first match of the pattern was going to
4930 fail. The "needed" character is now not set for anchored patterns, unless it
4931 follows something in the pattern that is of non-fixed length. Thus, it still
4932 fulfils its original purpose of finding quick non-matches in cases of nested
4933 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
4934
4935
4936 Version 4.0 17-Feb-03
4937 ---------------------
4938
4939 1. If a comment in an extended regex that started immediately after a meta-item
4940 extended to the end of string, PCRE compiled incorrect data. This could lead to
4941 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
4942
4943 2. Moved to autoconf 2.53 and libtool 1.4.2.
4944
4945 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
4946 the special perltest8 script is no longer needed - all the tests can be run
4947 from a single perltest script.
4948
4949 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
4950 by \s. It has now been removed in PCRE. This means it isn't recognized as
4951 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
4952 class [:space:] *does* include VT, thereby creating a mess.
4953
4954 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
4955 space and tab.
4956
4957 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
4958 its new features into the main test script, reducing the number of scripts.
4959
4960 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
4961 were backward compatible, and made the (?i) apply to the whole pattern, as if
4962 /i were given. Now it behaves more logically, and applies the option setting
4963 only to what follows. PCRE has been changed to follow suit. However, if it
4964 finds options settings right at the start of the pattern, it extracts them into
4965 the global options, as before. Thus, they show up in the info data.
4966
4967 8. Added support for the \Q...\E escape sequence. Characters in between are
4968 treated as literals. This is slightly different from Perl in that $ and @ are
4969 also handled as literals inside the quotes. In Perl, they will cause variable
4970 interpolation. Note the following examples:
4971
4972 Pattern PCRE matches Perl matches
4973
4974 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
4975 \Qabc\$xyz\E abc\$xyz abc\$xyz
4976 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4977
4978 For compatibility with Perl, \Q...\E sequences are recognized inside character
4979 classes as well as outside them.
4980
4981 9. Re-organized 3 code statements in pcretest to avoid "overflow in
4982 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
4983 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
4984 signed/unsigned warnings.
4985
4986 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
4987 option for pcretest, so I've replaced it by a simple function that does just
4988 that job.
4989
4990 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
4991 "pcregrep -".
4992
4993 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
4994 Java package. This provides some syntactic sugar for simple cases of what my
4995 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
4996 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
4997 item, you can use this simplified notation. Note that only makes sense with
4998 greedy quantifiers. Consequently, the use of the possessive quantifier forces
4999 greediness, whatever the setting of the PCRE_UNGREEDY option.
5000
5001 13. A change of greediness default within a pattern was not taking effect at
5002 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
5003 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
5004 was abstracted outside.
5005
5006 14. PCRE now supports the \G assertion. It is true when the current matching
5007 position is at the start point of the match. This differs from \A when the
5008 starting offset is non-zero. Used with the /g option of pcretest (or similar
5009 code), it works in the same way as it does for Perl's /g option. If all
5010 alternatives of a regex begin with \G, the expression is anchored to the start
5011 match position, and the "anchored" flag is set in the compiled expression.
5012
5013 15. Some bugs concerning the handling of certain option changes within patterns
5014 have been fixed. These applied to options other than (?ims). For example,
5015 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
5016 been the other way round. Some of this was related to change 7 above.
5017
5018 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
5019 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
5020 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
5021 POSIX classes only within a class (e.g. /[[:alpha:]]/).
5022
5023 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
5024 mode. Unlike ".", it always matches newline, whatever the setting of
5025 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
5026 assertions. Perl allows it, but it doesn't (in general) work because it can't
5027 calculate the length of the lookbehind. At least, that's the case for Perl
5028 5.8.0 - I've been told they are going to document that it doesn't work in
5029 future.
5030
5031 18. Added an error diagnosis for escapes that PCRE does not support: these are
5032 \L, \l, \N, \P, \p, \U, \u, and \X.
5033
5034 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
5035 reading past the end of the pattern in cases such as /[abcd/.
5036
5037 20. PCRE was getting more memory than necessary for patterns with classes that
5038 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
5039
5040 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
5041 compiling PCRE for use with Virtual Pascal.
5042
5043 22. Small fix to the Makefile to make it work properly if the build is done
5044 outside the source tree.
5045
5046 23. Added a new extension: a condition to go with recursion. If a conditional
5047 subpattern starts with (?(R) the "true" branch is used if recursion has
5048 happened, whereas the "false" branch is used only at the top level.
5049
5050 24. When there was a very long string of literal characters (over 255 bytes
5051 without UTF support, over 250 bytes with UTF support), the computation of how
5052 much memory was required could be incorrect, leading to segfaults or other
5053 strange effects.
5054
5055 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
5056 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
5057 there was a subsequent back reference to those brackets. This meant that, for
5058 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
5059 possible to check for precisely this case. All we can do is abandon the
5060 optimization if .* occurs inside capturing brackets when there are any back
5061 references whatsoever. (See below for a better fix that came later.)
5062
5063 26. The handling of the optimization for finding the first character of a
5064 non-anchored pattern, and for finding a character that is required later in the
5065 match were failing in some cases. This didn't break the matching; it just
5066 failed to optimize when it could. The way this is done has been re-implemented.
5067
5068 27. Fixed typo in error message for invalid (?R item (it said "(?p").
5069
5070 28. Added a new feature that provides some of the functionality that Perl
5071 provides with (?{...}). The facility is termed a "callout". The way it is done
5072 in PCRE is for the caller to provide an optional function, by setting
5073 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
5074 global variable. By default it is unset, which disables all calling out. To get
5075 the function called, the regex must include (?C) at appropriate points. This
5076 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
5077 This provides a means of identifying different callout points. When PCRE
5078 reaches such a point in the regex, if pcre_callout has been set, the external
5079 function is called. It is provided with data in a structure called
5080 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
5081 matching continues; if it returns a non-zero value, the match at the current
5082 point fails. However, backtracking will occur if possible. [This was changed
5083 later and other features added - see item 49 below.]
5084
5085 29. pcretest is upgraded to test the callout functionality. It provides a
5086 callout function that displays information. By default, it shows the start of
5087 the match and the current position in the text. There are some new data escapes
5088 to vary what happens:
5089
5090 \C+ in addition, show current contents of captured substrings
5091 \C- do not supply a callout function
5092 \C!n return 1 when callout number n is reached
5093 \C!n!m return 1 when callout number n is reached for the mth time
5094
5095 30. If pcregrep was called with the -l option and just a single file name, it
5096 output "<stdin>" if a match was found, instead of the file name.
5097
5098 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
5099 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
5100 pcre_exec(). This saves a malloc/free per call. The default value of
5101 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
5102 when configuring.
5103
5104 32. The default maximum size of a compiled pattern is 64K. There have been a
5105 few cases of people hitting this limit. The code now uses macros to handle the
5106 storing of links as offsets within the compiled pattern. It defaults to 2-byte
5107 links, but this can be changed to 3 or 4 bytes by --with-link-size when
5108 configuring. Tests 2 and 5 work only with 2-byte links because they output
5109 debugging information about compiled patterns.
5110
5111 33. Internal code re-arrangements:
5112
5113 (a) Moved the debugging function for printing out a compiled regex into
5114 its own source file (printint.c) and used #include to pull it into
5115 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
5116 separate copies.
5117
5118 (b) Defined the list of op-code names for debugging as a macro in
5119 internal.h so that it is next to the definition of the opcodes.
5120
5121 (c) Defined a table of op-code lengths for simpler skipping along compiled
5122 code. This is again a macro in internal.h so that it is next to the
5123 definition of the opcodes.
5124
5125 34. Added support for recursive calls to individual subpatterns, along the
5126 lines of Robin Houston's patch (but implemented somewhat differently).
5127
5128 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
5129 allow it to read and process whole directories in Win32. This code was
5130 contributed by Lionel Fourquaux; it has not been tested by me.
5131
5132 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
5133 used to name a group. Names consist of alphanumerics and underscores, and must
5134 be unique. Back references use the syntax (?P=name) and recursive calls use
5135 (?P>name) which is a PCRE extension to the Python extension. Groups still have
5136 numbers. The function pcre_fullinfo() can be used after compilation to extract
5137 a name/number map. There are three relevant calls:
5138
5139 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
5140 PCRE_INFO_NAMECOUNT yields the number of entries
5141 PCRE_INFO_NAMETABLE yields a pointer to the map.
5142
5143 The map is a vector of fixed-size entries. The size of each entry depends on
5144 the length of the longest name used. The first two bytes of each entry are the
5145 group number, most significant byte first. There follows the corresponding
5146 name, zero terminated. The names are in alphabetical order.
5147
5148 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
5149 case instead of 255. Making it the same both with and without UTF-8 support
5150 means that the same test output works with both.
5151
5152 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
5153 calling malloc() with a zero argument.
5154
5155 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
5156 optimization. I've improved things by keeping a bitmap of backreferences with
5157 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
5158 fact referenced, the optimization can be applied. It is unlikely that a
5159 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
5160 the match to follow \n) will appear inside brackets with a number greater than
5161 31, but if it does, any back reference > 31 suppresses the optimization.
5162
5163 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
5164 of disabling numbered capturing parentheses. Any opening parenthesis that is
5165 not followed by ? behaves as if it were followed by ?: but named parentheses
5166 can still be used for capturing (and they will acquire numbers in the usual
5167 way).
5168
5169 41. Redesigned the return codes from the match() function into yes/no/error so
5170 that errors can be passed back from deep inside the nested calls. A malloc
5171 failure while inside a recursive subpattern call now causes the
5172 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
5173
5174 42. It is now possible to set a limit on the number of times the match()
5175 function is called in a call to pcre_exec(). This facility makes it possible to
5176 limit the amount of recursion and backtracking, though not in a directly
5177 obvious way, because the match() function is used in a number of different
5178 circumstances. The count starts from zero for each position in the subject
5179 string (for non-anchored patterns). The default limit is, for compatibility, a
5180 large number, namely 10 000 000. You can change this in two ways:
5181
5182 (a) When configuring PCRE before making, you can use --with-match-limit=n
5183 to set a default value for the compiled library.
5184
5185 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
5186 a different value is set. See 45 below.
5187
5188 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
5189
5190 43. Added a new function pcre_config(int, void *) to enable run-time extraction
5191 of things that can be changed at compile time. The first argument specifies
5192 what is wanted and the second points to where the information is to be placed.
5193 The current list of available information is:
5194
5195 PCRE_CONFIG_UTF8
5196
5197 The output is an integer that is set to one if UTF-8 support is available;
5198 otherwise it is set to zero.
5199
5200 PCRE_CONFIG_NEWLINE
5201
5202 The output is an integer that it set to the value of the code that is used for
5203 newline. It is either LF (10) or CR (13).
5204
5205 PCRE_CONFIG_LINK_SIZE
5206
5207 The output is an integer that contains the number of bytes used for internal
5208 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
5209
5210 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
5211
5212 The output is an integer that contains the threshold above which the POSIX
5213 interface uses malloc() for output vectors. See item 31 above.
5214
5215 PCRE_CONFIG_MATCH_LIMIT
5216
5217 The output is an unsigned integer that contains the default limit of the number
5218 of match() calls in a pcre_exec() execution. See 42 above.
5219
5220 44. pcretest has been upgraded by the addition of the -C option. This causes it
5221 to extract all the available output from the new pcre_config() function, and to
5222 output it. The program then exits immediately.
5223
5224 45. A need has arisen to pass over additional data with calls to pcre_exec() in
5225 order to support additional features. One way would have been to define
5226 pcre_exec2() (for example) with extra arguments, but this would not have been
5227 extensible, and would also have required all calls to the original function to
5228 be mapped to the new one. Instead, I have chosen to extend the mechanism that
5229 is used for passing in "extra" data from pcre_study().
5230
5231 The pcre_extra structure is now exposed and defined in pcre.h. It currently
5232 contains the following fields:
5233
5234 flags a bitmap indicating which of the following fields are set
5235 study_data opaque data from pcre_study()
5236 match_limit a way of specifying a limit on match() calls for a specific
5237 call to pcre_exec()
5238 callout_data data for callouts (see 49 below)
5239
5240 The flag bits are also defined in pcre.h, and are
5241
5242 PCRE_EXTRA_STUDY_DATA
5243 PCRE_EXTRA_MATCH_LIMIT
5244 PCRE_EXTRA_CALLOUT_DATA
5245
5246 The pcre_study() function now returns one of these new pcre_extra blocks, with
5247 the actual study data pointed to by the study_data field, and the
5248 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
5249 before. That is, this change is entirely upwards-compatible and requires no
5250 change to existing code.
5251
5252 If you want to pass in additional data to pcre_exec(), you can either place it
5253 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
5254 block.
5255
5256 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
5257 data string contains the escape sequence \M, pcretest calls pcre_exec() several
5258 times with different match limits, until it finds the minimum value needed for
5259 pcre_exec() to complete. The value is then output. This can be instructive; for
5260 most simple matches the number is quite small, but for pathological cases it
5261 gets very large very quickly.
5262
5263 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
5264 returns the size of the data block pointed to by the study_data field in a
5265 pcre_extra block, that is, the value that was passed as the argument to
5266 pcre_malloc() when PCRE was getting memory in which to place the information
5267 created by pcre_study(). The fourth argument should point to a size_t variable.
5268 pcretest has been extended so that this information is shown after a successful
5269 pcre_study() call when information about the compiled regex is being displayed.
5270
5271 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
5272 because what follows is always an absolute path. (Later: it turns out that this
5273 is more than cosmetic for MinGW, because it doesn't like empty path
5274 components.)
5275
5276 49. Some changes have been made to the callout feature (see 28 above):
5277
5278 (i) A callout function now has three choices for what it returns:
5279
5280 0 => success, carry on matching
5281 > 0 => failure at this point, but backtrack if possible
5282 < 0 => serious error, return this value from pcre_exec()
5283
5284 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
5285 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
5286 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
5287 use by callout functions. It will never be used by PCRE itself.
5288
5289 (ii) The pcre_extra structure (see 45 above) has a void * field called
5290 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
5291 pcre_callout_block structure has a field of the same name. The contents of
5292 the field passed in the pcre_extra structure are passed to the callout
5293 function in the corresponding field in the callout block. This makes it
5294 easier to use the same callout-containing regex from multiple threads. For
5295 testing, the pcretest program has a new data escape
5296
5297 \C*n pass the number n (may be negative) as callout_data
5298
5299 If the callout function in pcretest receives a non-zero value as
5300 callout_data, it returns that value.
5301
5302 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
5303 there were some redundant $(CFLAGS) in commands that are now specified as
5304 $(LINK), which already includes $(CFLAGS).
5305
5306 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
5307 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
5308 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
5309 one-byte characters throughout. Note that case-insensitive matching applies
5310 only to characters whose values are less than 256. PCRE doesn't support the
5311 notion of cases for higher-valued characters.
5312
5313 (i) A character class whose characters are all within 0-255 is handled as
5314 a bit map, and the map is inverted for negative classes. Previously, a
5315 character > 255 always failed to match such a class; however it should
5316 match if the class was a negative one (e.g. [^ab]). This has been fixed.
5317
5318 (ii) A negated character class with a single character < 255 is coded as
5319 "not this character" (OP_NOT). This wasn't working properly when the test
5320 character was multibyte, either singly or repeated.
5321
5322 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
5323 mode, for example: \x{100}{2,3}.
5324
5325 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
5326 singly or repeated) now correctly test multibyte characters. However,
5327 PCRE doesn't recognize any characters with values greater than 255 as
5328 digits, spaces, or word characters. Such characters always match \D, \S,
5329 and \W, and never match \d, \s, or \w.
5330
5331 (v) Classes may now contain characters and character ranges with values
5332 greater than 255. For example: [ab\x{100}-\x{400}].
5333
5334 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
5335 PCRE in UTF-8 mode.
5336
5337 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
5338 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
5339 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
5340 value.)
5341
5342 53. The single man page has become too large. I have therefore split it up into
5343 a number of separate man pages. These also give rise to individual HTML pages;
5344 these are now put in a separate directory, and there is an index.html page that
5345 lists them all. Some hyperlinking between the pages has been installed.
5346
5347 54. Added convenience functions for handling named capturing parentheses.
5348
5349 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
5350 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
5351 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
5352 are faulted.
5353
5354 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
5355 calling configure. These values are used when compiling the dftables.c program
5356 which is run to generate the source of the default character tables. They
5357 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
5358 you will need to set these values.
5359
5360 57. Updated the building process for Windows DLL, as provided by Fred Cox.
5361
5362
5363 Version 3.9 02-Jan-02
5364 ---------------------
5365
5366 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
5367
5368 2. If --disable-static was given, the building process failed when trying to
5369 build pcretest and pcregrep. (For some reason it was using libtool to compile
5370 them, which is not right, as they aren't part of the library.)
5371
5372
5373 Version 3.8 18-Dec-01
5374 ---------------------
5375
5376 1. The experimental UTF-8 code was completely screwed up. It was packing the
5377 bytes in the wrong order. How dumb can you get?
5378
5379
5380 Version 3.7 29-Oct-01
5381 ---------------------
5382
5383 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
5384 This caused pcretest, when used on the test data, to segfault. Unfortunately,
5385 this didn't happen under Solaris 8, where I normally test things.
5386
5387 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
5388 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
5389 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
5390 this fix an hour or so after the initial 3.7 release.)
5391
5392
5393 Version 3.6 23-Oct-01
5394 ---------------------
5395
5396 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
5397 offsets passed as NULL with zero offset count.
5398
5399 2. The config.guess and config.sub files had not been updated when I moved to
5400 the latest autoconf.
5401
5402
5403 Version 3.5 15-Aug-01
5404 ---------------------
5405
5406 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
5407 had been forgotten.
5408
5409 2. By using declared but undefined structures, we can avoid using "void"
5410 definitions in pcre.h while keeping the internal definitions of the structures
5411 private.
5412
5413 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
5414 user point of view, this means that both static and shared libraries are built
5415 by default, but this can be individually controlled. More of the work of
5416 handling this static/shared cases is now inside libtool instead of PCRE's make
5417 file.
5418
5419 4. The pcretest utility is now installed along with pcregrep because it is
5420 useful for users (to test regexs) and by doing this, it automatically gets
5421 relinked by libtool. The documentation has been turned into a man page, so
5422 there are now .1, .txt, and .html versions in /doc.
5423
5424 5. Upgrades to pcregrep:
5425 (i) Added long-form option names like gnu grep.
5426 (ii) Added --help to list all options with an explanatory phrase.
5427 (iii) Added -r, --recursive to recurse into sub-directories.
5428 (iv) Added -f, --file to read patterns from a file.
5429
5430 6. pcre_exec() was referring to its "code" argument before testing that
5431 argument for NULL (and giving an error if it was NULL).
5432
5433 7. Upgraded Makefile.in to allow for compiling in a different directory from
5434 the source directory.
5435
5436 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
5437 options bits, the pointer it was passed was to an int instead of to an unsigned
5438 long int. This mattered only on 64-bit systems.
5439
5440 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
5441 generated) instead of pcre.in, which it its source. Also made the same change
5442 in several of the .c files.
5443
5444 10. A new release of gcc defines printf() as a macro, which broke pcretest
5445 because it had an ifdef in the middle of a string argument for printf(). Fixed
5446 by using separate calls to printf().
5447
5448 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
5449 script, to force use of CR or LF instead of \n in the source. On non-Unix
5450 systems, the value can be set in config.h.
5451
5452 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
5453 absolute limit. Changed the text of the error message to make this clear, and
5454 likewise updated the man page.
5455
5456 13. The limit of 99 on the number of capturing subpatterns has been removed.
5457 The new limit is 65535, which I hope will not be a "real" limit.
5458
5459
5460 Version 3.4 22-Aug-00
5461 ---------------------
5462
5463 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
5464
5465 2. Diagnose condition (?(0) as an error instead of crashing on matching.
5466
5467
5468 Version 3.3 01-Aug-00
5469 ---------------------
5470
5471 1. If an octal character was given, but the value was greater than \377, it
5472 was not getting masked to the least significant bits, as documented. This could
5473 lead to crashes in some systems.
5474
5475 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
5476 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
5477
5478 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
5479 These just pass their arguments on to (pcre_free)(), but they are provided
5480 because some uses of PCRE bind it to non-C systems that can call its functions,
5481 but cannot call free() or pcre_free() directly.
5482
5483 4. Add "make test" as a synonym for "make check". Corrected some comments in
5484 the Makefile.
5485
5486 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
5487 Makefile.
5488
5489 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
5490 command called pgrep for grepping around the active processes.
5491
5492 7. Added the beginnings of support for UTF-8 character strings.
5493
5494 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
5495 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
5496 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
5497 out for the ar command.)
5498
5499
5500 Version 3.2 12-May-00
5501 ---------------------
5502
5503 This is purely a bug fixing release.
5504
5505 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
5506 of ZA. This was just one example of several cases that could provoke this bug,
5507 which was introduced by change 9 of version 2.00. The code for breaking
5508 infinite loops after an iteration that matches an empty string was't working
5509 correctly.
5510
5511 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
5512 when matched against abbab (for example). After matching an empty string, it
5513 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
5514 caused it to match further down the string than it should.
5515
5516 3. The code contained an inclusion of sys/types.h. It isn't clear why this
5517 was there because it doesn't seem to be needed, and it causes trouble on some
5518 systems, as it is not a Standard C header. It has been removed.
5519
5520 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
5521 were reported on the Macintosh. The changes were from
5522
5523 while ((c = *(++ptr)) != 0 && c != '\n');
5524 to
5525 while ((c = *(++ptr)) != 0 && c != '\n') ;
5526
5527 Totally extraordinary, but if that's what it takes...
5528
5529 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
5530 available. Added HAVE_BCOPY and an autoconf test for it; if neither
5531 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
5532 assumes the way PCRE uses memmove() (always moving upwards).
5533
5534 6. PCRE is being used in one environment where strchr() is not available. There
5535 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
5536 faster code anyway.
5537
5538
5539 Version 3.1 09-Feb-00
5540 ---------------------
5541
5542 The only change in this release is the fixing of some bugs in Makefile.in for
5543 the "install" target:
5544
5545 (1) It was failing to install pcreposix.h.
5546
5547 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
5548
5549
5550 Version 3.0 01-Feb-00
5551 ---------------------
5552
5553 1. Add support for the /+ modifier to perltest (to output $` like it does in
5554 pcretest).
5555
5556 2. Add support for the /g modifier to perltest.
5557
5558 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
5559 matches null strings.
5560
5561 4. Fix perltest so that it doesn't do unwanted things when fed an empty
5562 pattern. Perl treats empty patterns specially - it reuses the most recent
5563 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
5564 effect.
5565
5566 5. The POSIX interface was broken in that it was just handing over the POSIX
5567 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
5568 required a bigger vector, with some working space on the end. This means that
5569 the POSIX wrapper now has to get and free some memory, and copy the results.
5570
5571 6. Added some simple autoconf support, placing the test data and the
5572 documentation in separate directories, re-organizing some of the
5573 information files, and making it build pcre-config (a GNU standard). Also added
5574 libtool support for building PCRE as a shared library, which is now the
5575 default.
5576
5577 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
5578 09 are not valid octal constants. Single digits will be used for minor values
5579 less than 10.
5580
5581 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
5582 existing programs that set these in the POSIX interface can use PCRE without
5583 modification.
5584
5585 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
5586 return all that pcre_info() returns, plus additional data. The pcre_info()
5587 function is retained for compatibility, but is considered to be obsolete.
5588
5589 10. Added experimental recursion feature (?R) to handle one common case that
5590 Perl 5.6 will be able to do with (?p{...}).
5591
5592 11. Added support for POSIX character classes like [:alpha:], which Perl is
5593 adopting.
5594
5595
5596 Version 2.08 31-Aug-99
5597 ----------------------
5598
5599 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
5600 trying to match at the startoffset position, but instead was moving forward to
5601 the next newline as if a previous match had failed.
5602
5603 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
5604 and could get into a loop if a null string was matched other than at the start
5605 of the subject.
5606
5607 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
5608 be distinguished at compile time, and for completeness also added PCRE_DATE.
5609
5610 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
5611 in GnuWin32 environments.
5612
5613
5614 Version 2.07 29-Jul-99
5615 ----------------------
5616
5617 1. The documentation is now supplied in plain text form and HTML as well as in
5618 the form of man page sources.
5619
5620 2. C++ compilers don't like assigning (void *) values to other pointer types.
5621 In particular this affects malloc(). Although there is no problem in Standard
5622 C, I've put in casts to keep C++ compilers happy.
5623
5624 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
5625 should be (const char *).
5626
5627 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
5628 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
5629 However, I haven't made this a standard facility. The documentation doesn't
5630 mention it, and the Makefile doesn't support it.
5631
5632 5. The Makefile now contains an "install" target, with editable destinations at
5633 the top of the file. The pcretest program is not installed.
5634
5635 6. pgrep -V now gives the PCRE version number and date.
5636
5637 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
5638 causing the entire string to be ignored, instead of just the last character.
5639
5640 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
5641 non-matching string, it can take a very, very long time, even for strings of
5642 quite modest length, because of the nested recursion. PCRE now does better in
5643 some of these cases. It does this by remembering the last required literal
5644 character in the pattern, and pre-searching the subject to ensure it is present
5645 before running the real match. In other words, it applies a heuristic to detect
5646 some types of certain failure quickly, and in the above example, if presented
5647 with a string that has no trailing " it gives "no match" very quickly.
5648
5649 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
5650 other alternatives are tried instead.
5651
5652
5653 Version 2.06 09-Jun-99
5654 ----------------------
5655
5656 1. Change pcretest's output for amount of store used to show just the code
5657 space, because the remainder (the data block) varies in size between 32-bit and
5658 64-bit systems.
5659
5660 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
5661 start matching at. This allows lookbehinds to work when searching for multiple
5662 occurrences in a string.
5663
5664 3. Added additional options to pcretest for testing multiple occurrences:
5665
5666 /+ outputs the rest of the string that follows a match
5667 /g loops for multiple occurrences, using the new startoffset argument
5668 /G loops for multiple occurrences by passing an incremented pointer
5669
5670 4. PCRE wasn't doing the "first character" optimization for patterns starting
5671 with \b or \B, though it was doing it for other lookbehind assertions. That is,
5672 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
5673 the letter 'x'. On long subject strings, this gives a significant speed-up.
5674
5675
5676 Version 2.05 21-Apr-99
5677 ----------------------
5678
5679 1. Changed the type of magic_number from int to long int so that it works
5680 properly on 16-bit systems.
5681
5682 2. Fixed a bug which caused patterns starting with .* not to work correctly
5683 when the subject string contained newline characters. PCRE was assuming
5684 anchoring for such patterns in all cases, which is not correct because .* will
5685 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
5686 DOTALL is set at top level; otherwise it knows that patterns starting with .*
5687 must be retried after every newline in the subject.
5688
5689
5690 Version 2.04 18-Feb-99
5691 ----------------------
5692
5693 1. For parenthesized subpatterns with repeats whose minimum was zero, the
5694 computation of the store needed to hold the pattern was incorrect (too large).
5695 If such patterns were nested a few deep, this could multiply and become a real
5696 problem.
5697
5698 2. Added /M option to pcretest to show the memory requirement of a specific
5699 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
5700
5701 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
5702 compiled in such a way that the backtracking after subsequent failure was
5703 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
5704 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
5705
5706
5707 Version 2.03 02-Feb-99
5708 ----------------------
5709
5710 1. Fixed typo and small mistake in man page.
5711
5712 2. Added 4th condition (GPL supersedes if conflict) and created separate
5713 LICENCE file containing the conditions.
5714
5715 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
5716 Perl, that is the internal \ allows the delimiter to be included in the
5717 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
5718 the final delimiter, add \ to the end of the pattern (to test the error).
5719
5720 4. Added the convenience functions for extracting substrings after a successful
5721 match. Updated pcretest to make it able to test these functions.
5722
5723
5724 Version 2.02 14-Jan-99
5725 ----------------------
5726
5727 1. Initialized the working variables associated with each extraction so that
5728 their saving and restoring doesn't refer to uninitialized store.
5729
5730 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
5731 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
5732 fix the problem.
5733
5734 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
5735 calls, and wasn't printing the correct value for compiling calls. Increased the
5736 default value of LOOPREPEAT, and the number of significant figures in the
5737 times.
5738
5739 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
5740
5741 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
5742 a building problem on Windows NT with a FAT file system.
5743
5744
5745 Version 2.01 21-Oct-98
5746 ----------------------
5747
5748 1. Changed the API for pcre_compile() to allow for the provision of a pointer
5749 to character tables built by pcre_maketables() in the current locale. If NULL
5750 is passed, the default tables are used.
5751
5752
5753 Version 2.00 24-Sep-98
5754 ----------------------
5755
5756 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
5757 it any more.
5758
5759 2. Allow quantification of (?>) groups, and make it work correctly.
5760
5761 3. The first character computation wasn't working for (?>) groups.
5762
5763 4. Correct the implementation of \Z (it is permitted to match on the \n at the
5764 end of the subject) and add 5.005's \z, which really does match only at the
5765 very end of the subject.
5766
5767 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
5768
5769 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
5770 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
5771 localized options. All options to pcre_study() were also removed.
5772
5773 7. Add other new features from 5.005:
5774
5775 $(?<= positive lookbehind
5776 $(?<! negative lookbehind
5777 (?imsx-imsx) added the unsetting capability
5778 such a setting is global if at outer level; local otherwise
5779 (?imsx-imsx:) non-capturing groups with option setting
5780 (?(cond)re|re) conditional pattern matching
5781
5782 A backreference to itself in a repeated group matches the previous
5783 captured string.
5784
5785 8. General tidying up of studying (both automatic and via "study")
5786 consequential on the addition of new assertions.
5787
5788 9. As in 5.005, unlimited repeated groups that could match an empty substring
5789 are no longer faulted at compile time. Instead, the loop is forcibly broken at
5790 runtime if any iteration does actually match an empty substring.
5791
5792 10. Include the RunTest script in the distribution.
5793
5794 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
5795 discrepancies, some of which were old and were also with respect to 5.004. They
5796 have now been fixed.
5797
5798
5799 Version 1.09 28-Apr-98
5800 ----------------------
5801
5802 1. A negated single character class followed by a quantifier with a minimum
5803 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
5804 program crashes, or just wrong answers. This did not apply to negated classes
5805 containing more than one character, or to minima other than one.
5806
5807
5808 Version 1.08 27-Mar-98
5809 ----------------------
5810
5811 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
5812
5813 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
5814 latter must appear before anything that relies on it in the pattern.
5815
5816
5817 Version 1.07 16-Feb-98
5818 ----------------------
5819
5820 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
5821 repeat of a potentially empty string).
5822
5823
5824 Version 1.06 23-Jan-98
5825 ----------------------
5826
5827 1. Added Markus Oberhumer's little patches for C++.
5828
5829 2. Literal strings longer than 255 characters were broken.
5830
5831
5832 Version 1.05 23-Dec-97
5833 ----------------------
5834
5835 1. Negated character classes containing more than one character were failing if
5836 PCRE_CASELESS was set at run time.
5837
5838
5839 Version 1.04 19-Dec-97
5840 ----------------------
5841
5842 1. Corrected the man page, where some "const" qualifiers had been omitted.
5843
5844 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
5845 input syntax.
5846
5847 3. Fixed memory leak which occurred when a regex with back references was
5848 matched with an offsets vector that wasn't big enough. The temporary memory
5849 that is used in this case wasn't being freed if the match failed.
5850
5851 4. Tidied pcretest to ensure it frees memory that it gets.
5852
5853 5. Temporary memory was being obtained in the case where the passed offsets
5854 vector was exactly big enough.
5855
5856 6. Corrected definition of offsetof() from change 5 below.
5857
5858 7. I had screwed up change 6 below and broken the rules for the use of
5859 setjmp(). Now fixed.
5860
5861
5862 Version 1.03 18-Dec-97
5863 ----------------------
5864
5865 1. A erroneous regex with a missing opening parenthesis was correctly
5866 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
5867 on some systems.
5868
5869 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
5870 it was reported that one broken compiler failed on the former because "code" is
5871 also an independent variable.
5872
5873 3. The erroneous regex a[]b caused an array overrun reference.
5874
5875 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
5876 fail on data ending with that character. (It was going on too far, and checking
5877 the next character, typically a binary zero.) This was specific to the
5878 optimized code for single-character negative classes.
5879
5880 5. Added a contributed patch from the TIN world which does the following:
5881
5882 + Add an undef for memmove, in case the the system defines a macro for it.
5883
5884 + Add a definition of offsetof(), in case there isn't one. (I don't know
5885 the reason behind this - offsetof() is part of the ANSI standard - but
5886 it does no harm).
5887
5888 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
5889 most of the places where whitespace preceded '#'. I have given up and
5890 allowed the remaining 2 cases to be at the margin.
5891
5892 + Rename some variables in pcre to eliminate shadowing. This seems very
5893 pedantic, but does no harm, of course.
5894
5895 6. Moved the call to setjmp() into its own function, to get rid of warnings
5896 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
5897
5898 7. Constructs such as \d{8,} were compiling into the equivalent of
5899 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
5900 outcome, but in this particular case used more store than had been allocated,
5901 which caused the bug to be discovered because it threw up an internal error.
5902
5903 8. The debugging code in both pcre and pcretest for outputting the compiled
5904 form of a regex was going wrong in the case of back references followed by
5905 curly-bracketed repeats.
5906
5907
5908 Version 1.02 12-Dec-97
5909 ----------------------
5910
5911 1. Typos in pcre.3 and comments in the source fixed.
5912
5913 2. Applied a contributed patch to get rid of places where it used to remove
5914 'const' from variables, and fixed some signed/unsigned and uninitialized
5915 variable warnings.
5916
5917 3. Added the "runtest" target to Makefile.
5918
5919 4. Set default compiler flag to -O2 rather than just -O.
5920
5921
5922 Version 1.01 19-Nov-97
5923 ----------------------
5924
5925 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
5926 like /([ab]*)*/, that is, for classes with more than one character in them.
5927
5928 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
5929 as /((?>a*))*/ (a PCRE_EXTRA facility).
5930
5931
5932 Version 1.00 18-Nov-97
5933 ----------------------
5934
5935 1. Added compile-time macros to support systems such as SunOS4 which don't have
5936 memmove() or strerror() but have other things that can be used instead.
5937
5938 2. Arranged that "make clean" removes the executables.
5939
5940
5941 Version 0.99 27-Oct-97
5942 ----------------------
5943
5944 1. Fixed bug in code for optimizing classes with only one character. It was
5945 initializing a 32-byte map regardless, which could cause it to run off the end
5946 of the memory it had got.
5947
5948 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
5949
5950
5951 Version 0.98 22-Oct-97
5952 ----------------------
5953
5954 1. Fixed bug in code for handling temporary memory usage when there are more
5955 back references than supplied space in the ovector. This could cause segfaults.
5956
5957
5958 Version 0.97 21-Oct-97
5959 ----------------------
5960
5961 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
5962
5963 2. Optimized negated single characters not to use a bit map.
5964
5965 3. Brought error texts together as macro definitions; clarified some of them;
5966 fixed one that was wrong - it said "range out of order" when it meant "invalid
5967 escape sequence".
5968
5969 4. Changed some char * arguments to const char *.
5970
5971 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
5972
5973 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
5974 pcretest.
5975
5976
5977 Version 0.96 16-Oct-97
5978 ----------------------
5979
5980 1. Added a simple "pgrep" utility to the distribution.
5981
5982 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
5983 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
5984 where "ddd" means "one or more decimal digits".
5985
5986 3. Fixed serious bug. If a pattern had a back reference, but the call to
5987 pcre_exec() didn't supply a large enough ovector to record the related
5988 identifying subpattern, the match always failed. PCRE now remembers the number
5989 of the largest back reference, and gets some temporary memory in which to save
5990 the offsets during matching if necessary, in order to ensure that
5991 backreferences always work.
5992
5993 4. Increased the compatibility with Perl in a number of ways:
5994
5995 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
5996 to request this handling. The option can be set at compile or exec time.
5997
5998 (b) $ matches before a terminating newline by default; an option
5999 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
6000 mode). The option can be set at compile or exec time.
6001
6002 (c) The handling of \ followed by a digit other than 0 is now supposed to be
6003 the same as Perl's. If the decimal number it represents is less than 10
6004 or there aren't that many previous left capturing parentheses, an octal
6005 escape is read. Inside a character class, it's always an octal escape,
6006 even if it is a single digit.
6007
6008 (d) An escaped but undefined alphabetic character is taken as a literal,
6009 unless PCRE_EXTRA is set. Currently this just reserves the remaining
6010 escapes.
6011
6012 (e) {0} is now permitted. (The previous item is removed from the compiled
6013 pattern).
6014
6015 5. Changed all the names of code files so that the basic parts are no longer
6016 than 10 characters, and abolished the teeny "globals.c" file.
6017
6018 6. Changed the handling of character classes; they are now done with a 32-byte
6019 bit map always.
6020
6021 7. Added the -d and /D options to pcretest to make it possible to look at the
6022 internals of compilation without having to recompile pcre.
6023
6024
6025 Version 0.95 23-Sep-97
6026 ----------------------
6027
6028 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
6029 \x20 at the start of a run of normal characters. These were being treated as
6030 real characters, instead of the source characters being re-checked.
6031
6032
6033 Version 0.94 18-Sep-97
6034 ----------------------
6035
6036 1. The functions are now thread-safe, with the caveat that the global variables
6037 containing pointers to malloc() and free() or alternative functions are the
6038 same for all threads.
6039
6040 2. Get pcre_study() to generate a bitmap of initial characters for non-
6041 anchored patterns when this is possible, and use it if passed to pcre_exec().
6042
6043
6044 Version 0.93 15-Sep-97
6045 ----------------------
6046
6047 1. /(b)|(:+)/ was computing an incorrect first character.
6048
6049 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
6050 but not actually doing anything yet.
6051
6052 3. Treat "-" characters in classes that cannot be part of ranges as literals,
6053 as Perl does (e.g. [-az] or [az-]).
6054
6055 4. Set the anchored flag if a branch starts with .* or .*? because that tests
6056 all possible positions.
6057
6058 5. Split up into different modules to avoid including unneeded functions in a
6059 compiled binary. However, compile and exec are still in one module. The "study"
6060 function is split off.
6061